WO2022048472A1 - 足式机器人运动控制方法、装置、设备及介质 - Google Patents

足式机器人运动控制方法、装置、设备及介质 Download PDF

Info

Publication number
WO2022048472A1
WO2022048472A1 PCT/CN2021/114272 CN2021114272W WO2022048472A1 WO 2022048472 A1 WO2022048472 A1 WO 2022048472A1 CN 2021114272 W CN2021114272 W CN 2021114272W WO 2022048472 A1 WO2022048472 A1 WO 2022048472A1
Authority
WO
WIPO (PCT)
Prior art keywords
centroid
target
path
foot
relationship
Prior art date
Application number
PCT/CN2021/114272
Other languages
English (en)
French (fr)
Inventor
郑宇�
姜鑫洋
迟万超
张正友
Original Assignee
腾讯科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 腾讯科技（深圳）有限公司 filed Critical 腾讯科技（深圳）有限公司
Priority to EP21863539.9A priority Critical patent/EP4116782A4/en
Publication of WO2022048472A1 publication Critical patent/WO2022048472A1/zh
Priority to US17/945,659 priority patent/US20230016514A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02—Control of position or course in two dimensions
    • G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0231—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means
    • G05D1/0246—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means using a video camera in combination with image processing means
    • G05D1/0251—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means using a video camera in combination with image processing means extracting 3D information from a plurality of images taken from different locations, e.g. stereo vision
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/18—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form
    • G05B19/19—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form characterised by positioning or contouring control systems, e.g. to control position from one programmed point to another or to control movement along a programmed continuous path
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B62—LAND VEHICLES FOR TRAVELLING OTHERWISE THAN ON RAILS
    • B62D—MOTOR VEHICLES; TRAILERS
    • B62D57/00—Vehicles characterised by having other propulsion or other ground- engaging means than wheels or endless track, alone or in addition to wheels or endless track
    • B62D57/02—Vehicles characterised by having other propulsion or other ground- engaging means than wheels or endless track, alone or in addition to wheels or endless track with ground-engaging propulsion means, e.g. walking members
    • B62D57/032—Vehicles characterised by having other propulsion or other ground- engaging means than wheels or endless track, alone or in addition to wheels or endless track with ground-engaging propulsion means, e.g. walking members with alternately or sequentially lifted supporting base and legs; with alternately or sequentially lifted feet or skid
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/35—Nc in input of data, input till input file format
    • G05B2219/35349—Display part, programmed locus and tool path, traject, dynamic locus

Definitions

  • the present application relates to the field of robotics, and in particular, to a motion control method, device, device and medium for a footed robot.
  • the position of the center of mass of the robot is solved according to the motion equation of the footed robot.
  • this method does not consider the actual motion process of the footed robot, so that the determined position of the center of mass does not conform to the actual motion of the footed robot.
  • the centroid positions generated in the technique have poor adaptability.
  • Embodiments of the present application provide a motion control method, device, device, and medium for a footed robot, which are used to improve the adaptability of the generated centroid position.
  • a motion control method for a footed robot is provided, which is executed by a motion control device for a footed robot, including:
  • the target centroid position variation coefficient and the target foothold that satisfy the variation relationship are screened; wherein the constraint condition set includes spatial foothold constraint conditions;
  • the footed robot is controlled to move according to the motion path.
  • a motion control device for a footed robot comprising:
  • the acquisition module is used to acquire the centroid state data corresponding to the space path start point and the space path end point of the motion path;
  • a first determination module configured to determine the candidate landing points of each foot end in the motion path based on the spatial path start point and the spatial path end point;
  • a second determination module configured to determine the variation relationship between the coefficient of change of the position of the center of mass and the contact force of the foot end based on the state data of the center of mass
  • a screening module configured to screen the target centroid position variation coefficient and target foothold that satisfy the variation relationship under the constraint of the constraint condition set; wherein the constraint condition set includes spatial foothold constraint conditions;
  • a third determining module configured to determine target motion control parameters according to the target centroid position variation coefficient and the target foothold;
  • the control module is configured to control the footed robot to move according to the movement path based on the target movement control parameter.
  • Embodiments of the present application provide a motion control device for a footed robot, including:
  • At least one processor connected to the memory
  • the memory stores instructions executable by the at least one processor, and the at least one processor implements the method according to any one of the aspects by executing the instructions stored in the memory.
  • An embodiment of the present application provides a storage medium, where computer instructions are stored in the storage medium, and when the computer instructions are executed on a computer, the computer is caused to execute the method according to any one of the aspects.
  • FIG. 1 is an example diagram of an application scenario of the motion control method for a footed robot provided by an embodiment of the present application
  • FIG. 2 is a structural example diagram of a control device provided by an embodiment of the present application.
  • FIG. 3 is an example diagram of a workspace provided by an embodiment of the present application.
  • FIG. 4 is a schematic structural diagram of a footed robot provided by an embodiment of the present application.
  • FIG. 5 is a flowchart of a motion control method for a footed robot provided by an embodiment of the present application
  • FIG. 6 is a schematic diagram of the distribution of sampling moments provided by an embodiment of the present application.
  • FIG. 7 is a schematic diagram of a centroid motion trajectory provided by an embodiment of the present application.
  • FIG. 8 is a schematic structural diagram of a motion control device for a footed robot provided by an embodiment of the application.
  • FIG. 9 is a schematic structural diagram of a motion control device for a footed robot according to an embodiment of the present application.
  • Robot It includes all kinds of machines that simulate human behavior or other creatures in thought (such as robot dogs, robot cats, etc.). Certain computer programs are also called robots in a broad sense. In contemporary industry, a robot refers to an artificial robot that can automatically perform tasks to replace or assist human work. It can be an electromechanical device, or controlled by a computer program or electronic circuit.
  • Footed robot refers to a robot with foot ends.
  • a footed robot can be configured with one or more foot ends, such as biped, quadruped or hexapod. Each foot end is correspondingly configured with one or more joints.
  • Motion path refers to the motion path that the footed robot needs to complete.
  • the length of the motion path can be arbitrary, and there is no specific limitation.
  • the movement time required for the footed robot to complete the motion path is regarded as a sampling period.
  • the movement duration includes multiple times, and when actually determining the position of the centroid, the time selected from the multiple times is called the sampling time.
  • the starting point of the space path also known as the starting position, which refers to the position where the robot starts to move along the motion path.
  • the time corresponding to the starting point of the space path can be called the starting time.
  • Space path end point refers to the position where the robot stops moving along the motion path, and the time corresponding to the space path end point can be called the termination time.
  • Centroid state data refers to the data used to describe the state change of the robot's centroid, including one or more of the robot's centroid position, centroid velocity, or centroid acceleration.
  • the position of the center of mass it is the center position of the mass of the robot, which is used to describe the position of the robot. The position of the center of mass of the robot will change under different motion states.
  • the centroid velocity can be obtained by taking the first derivative of the centroid position with respect to the time interval
  • the centroid acceleration can be obtained by taking the second derivative of the centroid position with respect to the time interval.
  • the position of the centroid at the starting point of the space path can be called the starting centroid position
  • the centroid velocity at the starting point of the space path can be called the starting centroid velocity
  • the centroid acceleration at the starting point of the space path can be called the starting centroid acceleration.
  • the centroid position of the end point of the space path can be called the end point centroid position
  • the centroid velocity of the end point of the space path can be called the end point centroid velocity
  • the centroid acceleration of the end point of the space path can be called the end point centroid acceleration.
  • Candidate landing point refers to the possible contact position of the foot end and the contact surface when the foot end of the robot is in contact with the contact surface.
  • the candidate landing points of the robot include three points A, B, and C.
  • Variation coefficient of centroid position The coefficient of variation of centroid position is used to describe the parameter of centroid position variation in the process of time change.
  • the coefficient of change of the centroid position is expressed in the form of a matrix, or in the form of a vector, etc.
  • the centroid position variation coefficient and the time interval can jointly represent the centroid position at a specific time, and the time interval refers to the time difference between the specific time and the time corresponding to the starting point of the spatial path.
  • Contact surface refers to the surface that the robot's foot end contacts with the environment, such as the ground, or other supports that are in contact with the foot end. It should be noted that due to the uneven road surface, the contact surfaces corresponding to the multiple foot ends of the footed robot may be different.
  • Movement trajectory of the center of mass It can also be called the movement trajectory of the position of the center of mass, which is used to describe the position of the center of mass of the robot at different times. It actually consists of the position of the center of mass of the robot at different times.
  • Contact point It can also be called the foot end of the foot, which refers to the foot end of the robot in contact with the contact surface.
  • the number of the foot end in contact with the contact surface is the number of contact points.
  • the foot ends of the robot that are in contact with the contact surface are not exactly the same at different times.
  • Step sequence It is used to describe the gait of the robot in the process of completing the motion path, including at least one motion stage in the process of the robot moving along the motion path, and the duration of each motion stage.
  • Foot contact force refers to the force between the foot end and the contact surface of the robot in contact with the contact surface.
  • Constraint condition set refers to the relational expression used to constrain the value of the centroid position variation coefficient and/or the foothold point.
  • a constraint set includes one or more constraints.
  • the constraint condition set in this embodiment of the present application includes at least a spatial foothold constraint.
  • the spatial foothold constraint is used to constrain the foothold of the robot to be located in the working space of the robot, and the spatial foothold constraint is specifically related to the robot's center of mass. Position variation coefficient.
  • Target motion control parameters parameters required to control the motion of the footed robot, including the joint angle of the footed robot at any time, and the joint torque at any time.
  • multiple refers to two or more
  • at least one refers to one or more
  • a and/or B specifically includes A, B, and A and B. condition.
  • an embodiment of the present application provides a motion control method for a footed robot.
  • the following describes the design idea of the motion control method for a footed robot involved in the embodiment of the present application.
  • a plurality of candidate landing points corresponding to the foot end of the footed robot are first determined, and the relationship between the change coefficient of the center of mass position and the contact force of the foot end is determined according to the state data of the centroid of the footed robot, and based on the plurality of candidate landing points, Using the change relationship and constraint condition set, determine the target centroid position variation coefficient and target foothold of the footed robot, and then determine the centroid movement trajectory of the footed robot according to the target centroid position variation coefficient, and according to the centroid movement trajectory and the target foothold, Determine the target motion control parameters, and use the target motion control parameters to control the footed robot to move along the motion path.
  • the determined position of the centroid and the target foothold meet the constraints of the space foothold, which improves the adaptability between the determined position of the centroid and the actual motion process sex.
  • the method realizes the automatic generation of the position of the centroid and the motion trajectory of the centroid, and the automatic selection of the foothold, which improves the automation degree of controlling the footed robot.
  • a quadratic program can be constructed according to the variation relationship, the constraint condition set, and the objective function, and the variation coefficient of the target centroid position and the target foothold can be solved by solving the quadratic program. If there is a solution to the quadratic programming problem, the global optimal solution must be solved, and the change coefficient of the target centroid position and the target foothold are converted into quadratic programming problems, so that the optimal target centroid position change coefficient and target can be solved. foothold.
  • the motion control method of the footed robot is suitable for controlling various gaits of various kinds of footed robots in various environments.
  • Various types of footed robots such as biped robots, quadruped robots, etc.
  • Various environments such as flat ground, uneven ground, slopes or stairs, etc.
  • Various gaits such as bipedal walking, quadrupedal walking, quadrupedal trotting, and random gaits, etc.
  • FIG. 1 is an application scenario diagram of a motion control method for a footed robot, or can be understood as an architecture diagram of a footed robot motion control system.
  • the architecture diagram includes a footed robot 110 and a control device 120. The following describes the control An example of the interaction between the device 120 and the footed robot 110 is presented.
  • control device 120 and the footed robot 110 are two relatively independent devices:
  • the communication between the footed robot 110 and the control device 120 is performed by wire or wireless.
  • the communication between the footed robot 110 and the control device 120 is implemented using a communication network as an example.
  • the control device 120 may set the state data of the centroid of the space path starting point and the space path end point of the footed robot 110 according to the user's operation or the task of the footed robot 110 .
  • the footed robot 110 may detect the state data of the center of mass of the starting point of the space path, and upload the state data of the center of mass of the starting point of the space path to the control device 120 .
  • the control device 120 directly collects the state data of the centroid of the footed robot 110 at the starting point of the spatial path.
  • control device 120 collects an image of the environment where the footed robot 110 is currently located, or receives an environment image reported by the footed robot 110, and the control device 120 determines the foot end of the footed robot 110 that needs to rest in the motion path according to the environment image. Possible candidate footholds. Of course, there are many ways for the control device 120 to determine the candidate landing point, which will be described in detail below.
  • the control device 120 determines the movement trajectory of the centroid and the foothold of the footed robot 110 according to the state data of the centroid and the candidate landing points, and then controls the footed robot 110 to perform corresponding movements. Among them, the content of determining the motion trajectory and foothold of the centroid will be introduced below.
  • the control device 120 can be implemented by a server or a terminal.
  • the server includes but is not limited to: an independent physical server, a server cluster composed of multiple physical servers, or a distributed system, and can also provide cloud services, cloud databases, cloud computing, and cloud functions. , cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and cloud servers for basic cloud computing services such as big data and artificial intelligence platforms.
  • Terminals such as mobile phones, personal computers, smart TVs or portable tablet computers, etc.
  • control device 120 is part of the footed robot 110:
  • the control device 120 may be disposed in the body of the footed robot 110 , for example, the control device 120 is an internal processor in the footed robot 110 and the like.
  • the control device 120 may receive motion instructions from the host computer, or operate according to the user's input to obtain motion instructions, the motion instructions may instruct the footed robot 110 to perform a certain task, or instruct the footed robot 110 to perform a certain task.
  • the upper computer may be any device that is wirelessly or wiredly connected to the control device 120 , such as a terminal or a server.
  • the control device 120 collects the state data of the centroid of the footed robot 110 .
  • the control device 120 may determine a possible candidate footing point for the foot end of the footed robot 110 each time according to the image of the environment where the footed robot 110 is currently located.
  • the control device 120 determines the movement trajectory of the centroid of the footed robot and the target foothold according to the centroid state data and the candidate footholds, so as to control the footed robot 110 to move.
  • the content of determining the motion trajectory and foothold of the centroid will be introduced below.
  • the control device 120 includes a visual perception unit 210 , a trajectory generation unit 220 and a motion control unit. 230 for example.
  • the visual perception unit 210 may be disposed on the footed robot 110 , for example, mounted on the head of the footed robot 110 .
  • the visual perception unit 210 is, for example, one or more of a camera and an infrared camera. Camera such as RGBD camera.
  • the visual perception unit 210 collects the robot state.
  • the robot state includes the state data of the center of mass of the start point and the end point of the space path of the footed robot 110 .
  • the visual perception unit 210 may also collect an image of the environment where the footed robot 110 is located, and obtain possible candidate landing points for each footing of the footed robot 110 .
  • the step sequence of the footed robot 110 can also be obtained according to the motion path of the footed robot 110 and the like.
  • the visual perception unit 210 may send these data to the trajectory generation unit 220 .
  • the trajectory generation unit 220 receives and determines the position of the centroid and the target foothold of the footed robot 110 at multiple times according to the robot state data, the sequence of steps and the candidate landing points, and obtains the centroid of the footed robot 110 according to the position of the centroid at multiple times. movement trajectory. Then, the whole body movement trajectory of the footed robot 110 is determined according to the movement trajectory of the center of mass and the target foothold, and the whole body movement trajectory and the target foothold are sent to the motion control unit 230 .
  • the motion control unit 230 can determine the joint torque of each joint of the footed robot 110 according to the whole body motion trajectory and the target foothold, and control the rotation of each joint of the footed robot 110 according to the joint torque, thereby realizing the motion of the footed robot 110 .
  • the motion control unit 230 can also monitor the real-time state of the footed robot during the movement of the footed robot 110 to ensure that the footed robot 110 can move stably.
  • the embodiment of the present application converts the centroid position of the footed robot 110 at each moment into the relationship between the centroid position variation coefficient, the initial centroid position, and the time interval between the moment and the initial centroid position. According to the state data of the center of mass, the relationship between the position of the center of mass and the contact force of the foot end is converted into the relationship between the change coefficient of the center of mass position and the contact force of the foot end, and based on the change relationship and the set of constraints, the change of the center of mass position is solved.
  • the coefficient and the target foothold and then according to the change coefficient of the position of the centroid, obtain the motion trajectory of the foot robot 110, and determine the target motion control parameters of the foot robot 110 according to the motion trajectory of the centroid and the target foothold, and use the target motion control parameters to control the foot movement of the robot 110.
  • the problem of determining the coefficient of change of the centroid position and the target foothold can be converted into a mixed integer quadratic programming problem by combining the set of constraints and the change relationship. Secondary planning problem to obtain centroid locations and target footholds.
  • the change relationship and the set of constraints are pre-configured in the control device 120 , or acquired by the control device 120 from other devices or network resources, or created by the control device 120 . The following describes an example of the manner in which the control device 120 creates the change relationship and the constraint condition set:
  • the control device 120 can obtain the relationship between the position of the centroid at each moment and the foot contact force at the corresponding moment from network resources or other devices.
  • the position of the centroid at each moment in the relational expression is expressed by the initial centroid position, the coefficient of change of the centroid position and the time interval, so that the relational expression between the centroid position at each moment and the contact force of the foot end at the corresponding moment is converted into The relationship between the coefficient of change of the center of mass position and the contact force at the foot end.
  • the change relationship is used to represent the change relationship between the coefficient of change of the center of mass position and the contact force of the foot end.
  • the control device 120 may store the change relationship in any form, such as a function form, a description statement form, and the like.
  • the first relational expression refers to the center-of-mass dynamics equation of the footed robot 110 , and is used to express the relationship between the motion of the footed robot 110 and the external force it receives.
  • the first relation can have various expressions, such as the Newton-Euler equation.
  • An example of a first relation is as follows:
  • m is the total mass of the footed robot 110
  • g ⁇ R3 is the acceleration of gravity
  • pG ⁇ R3 is the position of the center of mass of the footed robot 110
  • pG ⁇ R3 is the position of the center of mass of the footed robot 110
  • L ⁇ R3 is the angular momentum of the centroid of the footed robot 110
  • f i ⁇ R 3 is the foot contact force of the i-th contact point
  • N c is the number of contact points
  • the operation refers to the diagonal diagonal matrix of (), Indicates the second derivative of the time interval by p G.
  • R 3 represents the three coordinate values in the coordinate system.
  • each quantity is a representation result in the world coordinate system.
  • each variable in formula (1) is the representation result in the world coordinate system.
  • G i ⁇ R 3 ⁇ 3 may be a 3 ⁇ 3 matrix.
  • p init represents the initial centroid position
  • p t represents the change of the centroid after the time interval t.
  • the first item and is linearly related to p t
  • the second term is a constant term
  • the third item can be determined according to the pre-configured body posture of the footed robot 110 at each moment value of .
  • pt pt xy + pt z
  • pt xy includes the components of pt on the x and y axes. It refers to the second derivative of the component of the centroid position change p t on the plane formed by the xy axis to the time interval, and also represents the component of the centroid acceleration on the plane formed by the x and y axes; It refers to the second derivative of the component of the centroid position change p t on the plane formed by the z-axis to the time interval, and also represents the component of the centroid acceleration on the plane formed by the z-axis; is an oblique diagonal matrix representing the xy-axis component of the centroid position change p t , A diagonal diagonal matrix representing the z-axis component of the centroid position change pt .
  • p t z includes the component of p t on the z axis, and the corresponding x and y coordinates of p t z are 0.
  • the moment generated about the z-axis is The moment in a certain direction in the xy plane is due to and collinear, so
  • the motion of the footed robot 110 in the z-axis direction is usually relatively stable, so it can be ignored as well as also, as well as The absolute value of is small and can also be ignored.
  • the neglected term of the above formula (6) that is, the moment involved in the formula (6), can also be obtained by adjusting the foot end force between the foot end and the contact surface. .
  • H 0 and w are known quantities
  • x t includes the change in the position of the center of mass p t and the acceleration of the center of mass to be quantified.
  • T P [1t...t n ] ⁇ R 1 ⁇ (n+1)
  • c * [c *,0 c *,1 ...c *,n ]
  • T ⁇ R n+1 is the polynomial coefficient
  • t represents the time interval, that is, the time interval between the moment and the moment corresponding to the starting point of the spatial path
  • * represents x, y and z
  • c refers to the coefficient of change of the centroid position, which includes all polynomial coefficients.
  • the centroid position at each moment in a plurality of moments can be calculated according to the above formula (8).
  • n is any integer greater than or equal to 2.
  • centroid acceleration is specifically expressed as follows:
  • Formula (10) represents the relationship between the center of mass position variation coefficient c and the foot contact force f i , which is actually obtained by transforming the relationship between the center of mass position and the foot contact force.
  • the selected candidate landing point refers to a landing point determined to be used as a foot end from a plurality of candidate landing points, and may also be regarded as a target landing point.
  • the foot of the footed robot 110 that needs to land usually has more than one candidate landing point, that is, ri i ⁇ r ij
  • j 1,2,3...N i ⁇ , which is involved in the foregoing, represents the i -th foot.
  • the position of the candidate landing point, N i represents the number of the candidate landing point of the i-th foot.
  • Formula (12) indicates that only one of the binary variables in Ni is equal to 1, and the rest are 0, which means that only one candidate landing point can be selected as the target landing point. Therefore, the selected candidate landing point can be introduced into the above-mentioned fifth relational formula to obtain The equation of motion of the center of mass rewritten as follows is:
  • k represents the kth moment
  • f ijk represents the foot contact force at the kth moment and the jth candidate footing point is selected by the ith foot.
  • H k represents the H corresponding to the kth moment, and the specific representation of H can refer to the foregoing.
  • w k represents w corresponding to the kth moment, and the specific representation of w can refer to the previous content.
  • the constraint condition set includes one or more constraint conditions, each constraint condition is a value for constraining one or both of the position of the centroid or the target landing point, and the form of each constraint condition may be an inequality.
  • the constraint condition set at least includes a space foothold constraint condition, and the space foothold constraint condition is used to constrain that the footholds corresponding to the position of the centroid of the footed robot 110 are all reachable by the foot end of the footed robot.
  • the control device 120 may obtain a set of constraints from network resources or other devices, or a set of constraints created by itself. The following will illustrate how the control device 120 creates each constraint:
  • xi represents the footing of the footed robot 110 in this coordinate system point
  • d il is the distance between the mid-surface of the convex polyhedron and the origin
  • s il represents the unit normal vector corresponding to the mid-surface of the convex polyhedron
  • l represents the number of surfaces corresponding to the convex polyhedron
  • c represents the coefficient of change of the centroid position
  • ⁇ ij represents The target landing point and the convex polyhedron are determined according to the bending range of the joint configured at the foot end of the foot-mounted robot 110 and the length of the joint.
  • the control device 120 discretely determines the reachable position of the foot end of the footed robot 110 relative to the joint according to the bending range of the joint and the length of the joint.
  • the joint is a joint configured at the foot end of the foot, and may specifically be a joint directly connected with the foot end, or other joints connected with the foot end through a joint.
  • the bending range of a joint refers to a range consisting of a minimum angle and a maximum angle that the joint can bend.
  • the bending range of the joint is 0° to 120°, and the bending range is generally known.
  • the length of the joint is, for example, 1 meter.
  • the control device 120 After discretely determining a plurality of positions that the foot end of the footed robot 110 can reach relative to the joint, the control device 120 fits the plurality of positions, so as to obtain the working space of the footed robot 110 .
  • the workspace of the foot end of the footed robot 110 is generally a non-convex region, but the workspace can be approximated as a convex polyhedron by fitting.
  • the convex polyhedron is specifically the convex polyhedron 310 shown in FIG. 3 , and each point of the plurality of points shown in FIG. 3 represents the reachable position of the foot end relative to the joint.
  • control device 120 can obtain a linear inequality representation of each face in the convex polyhedron, and the linear inequality representation of each face is specifically expressed as sil T x i ⁇ d il .
  • the control device 120 combines the inequalities of the respective faces of the convex polyhedron to obtain the formula (15) as described above.
  • the above formula (15) can be carried out in the local fixed coordinate system of the joint.
  • the local fixed coordinate system of the joint refers to the coordinate system established with the local part of the joint as the coordinate origin, that is, the local fixed coordinate system of the joint.
  • the coordinate origin may be different from the coordinate origin in the world coordinate system.
  • the joint may be any joint associated with the end of the foot on which the foot falls.
  • FIG. 4 is a schematic diagram of the structure of the footed robot 110 .
  • the footed robot 110 includes a plurality of joints 410 and four foot ends 420 , each foot end 420 is configured with a plurality of joints, and the local fixed coordinate system of the joints A coordinate system as shown in Figure 4.
  • each variable can be decomposed into a plurality of variables according to the local fixed coordinate system of the joint.
  • the foot contact force f i in FIG. 4 can be decomposed into ni , f i1 and f i2 along the local fixed coordinate system, and the centroid position p G can also be decomposed along the local fixed coordinate system.
  • the control device 120 has determined in advance the candidate landing point 430 corresponding to the foot that needs to be placed, and the control device 120 can determine whether a selected candidate landing point 430 is located on the foot according to the above formula (15). Inside the convex polyhedron. For example, it can be determined whether the footing point of the i-th foot end of the footed robot 110 is located in the convex polyhedron after the time interval t.
  • R 0 ⁇ R 3 ⁇ 3 is a rotation matrix, representing the body posture of the footed robot 110 at the current moment
  • p il ⁇ R 3 is the joint of the i-th foot in the footed robot 110 relative to the centroid of the footed robot 110 in the body
  • the position in the coordinate system, R il ⁇ R 3 is the body pose of the joint in the locally fixed coordinate system relative to the body coordinate system.
  • Both p il and R il are constants.
  • R 0 may be a constant, or the control device 120 may determine it according to the body posture of the footed robot 110 at the current moment.
  • the control device 120 combines the above formulas (4), (8), (16) and (15) to obtain the eighth relational formula:
  • a ik , Bi ik and bi ik represent A i , Bi and bi corresponding to the kth moment, respectively.
  • the constraint set may further include a friction constraint for constraining the magnitude of the frictional force between the foot end and the contact surface in contact with the contact surface, or a friction constraint for constraining the foot end contact force
  • a friction constraint for constraining the magnitude of the frictional force between the foot end and the contact surface in contact with the contact surface
  • a friction constraint for constraining the foot end contact force One or both of the contact force constraints where the contact force in the normal direction is less than or equal to the upper limit of the contact force.
  • the frictional force between the foot end of the foot and the contact surface is determined according to the contact force of the foot end and the friction coefficient.
  • the friction force between the foot end and the contact surface is also different at each moment, so the friction force constraint condition is actually different.
  • the size of the friction force between the foot end contacting the contact surface and the contact surface at each moment is constrained, and the contact force constraint condition is the size of the contact force in the normal direction of the foot end contact force at each moment.
  • control device 120 creates the friction force constraints and the contact force constraints:
  • N i -[ ⁇ i n i -o i ⁇ i n i +o i ⁇ i n i -t i ⁇ i n i +t i ] ⁇ R 3 ⁇ 4
  • n i represents the foot of the i-th foot
  • the normal vector of the point, o i represents the vector in one tangential direction of the foot end of the ith foot, t i is the vector in the other tangent direction of the foot end of the ith foot, ⁇ i represents the distance between the foot and the contact surface. friction coefficient between.
  • the corresponding friction coefficient values may also be different.
  • the coefficient of friction may also be different.
  • N ij represents the value of N i corresponding to the j-th candidate landing point selected by the i-th foot end .
  • f ijk represents the contact force of the foot end when the ith foot end selects the j th candidate footing point at the k th moment.
  • contact force constraints can be set to constrain the footed robot 110 and the contact surface.
  • the size of the contact force between the foot ends between the feet avoids the excessive force between each movement of the footed robot 110 and the contact surface.
  • f ijk represents the foot contact force at the k th moment when the j th candidate footing point is selected by the i th foot.
  • the target centroid position change coefficient and the target foothold can be randomly determined from the values satisfying the change relationship and the set of constraints.
  • an objective function may also be introduced in the embodiment of the present application, and the objective function is used to further screen out the optimal centroid position variation coefficient and foothold, and then the optimal centroid The position variation coefficient is determined as the target centroid position variation coefficient, and the optimal foothold is determined as the target foothold.
  • the determination of the target centroid position variation coefficient and the target foothold can be transformed into a quadratic programming problem. Therefore, the target in the embodiment of the present application
  • the function includes at least a quadratic term of one or more variables, and the one or more variables may be any variable that satisfies the change relationship and the candidate result related to the set of constraints.
  • Quadratic terms can be constructed from the quadratic power of a variable.
  • the objective function includes at least one of the following A1 to A5:
  • A1 The quadratic term related to the variation of the contact force of the foot end in the motion path
  • A2 The quadratic term of the change in the position of the centroid in the motion path
  • A3 The quadratic term of the difference between the end centroid position of the end point of the spatial path in the centroid state data and the centroid position of the end point of the space path in the candidate result;
  • A4 The quadratic term of the difference between the terminal centroid velocity at the end point of the spatial path in the centroid state data and the centroid velocity at the end point of the space path in the candidate result;
  • A5 The quadratic term of the difference between the end centroid acceleration at the end point of the space path in the centroid state data and the centroid acceleration at the end point of the space path in the candidate result.
  • Function of A1 The variation of the contact force of the foot end can be used to optimize the distribution of the force between the foot end and the contact surface, so that the distribution of the force between the foot end and the contact surface of the foot robot 110 during the walking process is more evenly.
  • the role of A2 The change in the position of the center of mass reflects the length of the motion trajectory of the center of mass, which is beneficial to reduce the oscillation amplitude of the motion trajectory of the center of mass.
  • J grf is the weighted sum of squares of all foot contact forces in the motion path
  • J len is the weighted sum of squares of the difference between the changes of the centroid position between every two adjacent moments
  • J tgt is the centroid state
  • the following describes how to use the constructed variation relationship and constraint condition set to determine the centroid in the embodiment of the present application in conjunction with the flow of the motion control method for a footed robot shown in FIG. 5 .
  • the process of the position and the target landing point is introduced as an example, please refer to FIG. 5 , the method can be executed by the motion control device of the footed robot shown in FIG. 9 , including:
  • the control device 120 controls the lifting and lowering of the foot end of the footed robot 110, thereby realizing the movement of the footed robot 110, so that the footed robot 110 can complete the path from the starting point of the space path to the space path
  • the starting point of the spatial path is the current position of the footed robot 110 .
  • the end point of the spatial path is the position to be reached by the footed robot 110 , and the end point of the spatial path may be set in advance, or determined by the control device 120 according to the task that the footed robot 110 needs to perform.
  • centroid state data may include centroid locations.
  • control device 120 may obtain the other two values according to one of the position of the centroid, the velocity of the centroid or the acceleration of the centroid.
  • centroid position, centroid velocity, or centroid acceleration can all be represented by coordinates in a coordinate system or by vectors, etc.
  • the motion path, the start point of the space path and the end point of the space path are related to the sampling period of the selected footed robot 110, and the motion path, the start point of the space path and the end point of the space path can be flexibly set according to actual needs .
  • control device 120 may determine the position corresponding to the current moment of the footed robot 110 as the starting point of the spatial path, the control device 120 may determine the position corresponding to the 3 s as the ending point of the spatial path, and in the next sampling period, the 3 s may be selected as the starting point of the spatial path.
  • the footed robot 110 may have one or more foot ends, and each foot end may land one or more times, which is specifically related to the set start point of the space path and the end point of the space path.
  • the control device 120 may pre-determine a plurality of candidate landing points for the foot end that needs to land each time.
  • the candidate landing points refer to possible landing points of the footed robot 110 in the motion path.
  • the landing point can be represented by coordinates or vectors in the world coordinate system.
  • control device 120 may collect an environment image of the footed robot 110 through the visual perception unit 210, and construct a conversion relationship between each pixel in the environment image and the world coordinate system. The control device 120 determines possible candidate landing points along the movement direction from the start point of the space path to the end point of the space path through the environment image and the conversion relationship.
  • control device 120 can identify obstacles from the starting point of the spatial path to the ending point of the spatial path according to the environment image, and determine the position of the non-obstacle along the moving direction corresponding to the starting point of the spatial path to the ending point of the spatial path according to the conversion relationship, and Use the identified location as a candidate landing spot.
  • the control device 120 collects a three-dimensional point cloud image of the environment through the visual perception unit 210, and determines a candidate landing point from the three-dimensional point cloud image according to the three-dimensional point cloud image of the environment. Specifically, the position of a non-obstacle may be determined from the three-dimensional point cloud image as a candidate landing point. point.
  • the visual perception unit 210 is an RGBD camera
  • a 3D point cloud image can be collected, or, for example, by collecting multiple environment images where the footed robot 120 is currently located, and performing 3D reconstruction on the multiple environment images to obtain a 3D point cloud image.
  • control device 120 may randomly select multiple positions as the candidate landing points in the moving direction from the start point of the spatial path to the end point of the spatial path.
  • the control device 110 can determine all possible candidate landing points of the robot in the movement path, and the all possible candidate landing points are the candidate landing points corresponding to each landing.
  • a large area where the footed robot 110 may land may be determined according to the motion speed of the footed robot 110 , and the candidate landings can be selected from the large areas corresponding to each landing in turn according to any of the above methods. point.
  • the control device 120 respectively determines that the candidate landing points 430 of the foot ends include a plurality of circles on the ground shown in FIG. 4 .
  • the change relationship is, for example, the formula (14) discussed above, and the meaning of the change relationship can be referred to the content discussed above, which will not be repeated here.
  • the middle control device 120 may obtain the centroid position at each of the multiple moments in the motion path, and then determine the centroid movement trajectory based on the centroid positions at the multiple moments.
  • the change relationship includes parameters such as H k and w k in addition to the center of mass position change coefficient and the contact force of the foot, so these parameters can be determined by known quantities such as the center of mass state data. The value of , and then obtain the change relationship that only includes the change coefficient of the center of mass position and the contact force of the foot end.
  • the control device 120 can obtain the moving time required for the footed robot to complete the motion path according to the length of the motion path and the motion speed of the footed robot 110 , or the control device 120 is directly pre-configured with the motion required for the footed robot 110 to complete the motion path. duration.
  • control device 120 may randomly sample from the movement duration to obtain multiple sampling moments. Random sampling to obtain the sampling moment is simpler.
  • control device 120 separately samples from the duration of each motion stage according to the stepping sequence of the footed robot 110 to obtain a plurality of sampling moments. Since each motion stage has its corresponding sampling time, it can be ensured that each motion stage can have a corresponding sampling time, which is beneficial to improve the accuracy of the center of mass motion trajectory determined later.
  • the time interval between any two adjacent sampling moments may be the same or different. Different means that the time interval between any two adjacent sampling moments is not exactly the same. , or the time interval between two adjacent sampling moments is different.
  • the sampling time at least includes the start time and the end time of each motion stage, and at least one intermediate time in each motion stage.
  • the intermediate moment refers to any moment between the start moment and the end moment of the motion phase.
  • the quadruped walking gait of the footed robot 110 is set as a sampling period, and the control device 120 sequentially divides the motion process of the footed robot 110 into eight motion stages in the sampling period.
  • the eight motion stages are specifically: four Foot support moving center of mass (referred to as 4S), step right hind foot (referred to as HR), step right front foot (referred to as FR), quadruped support moving center of mass (referred to as 4S), quadruped support moving center of mass (referred to as 4S) , Mai left hind foot (referred to as HL), Mai left front foot (referred to as FL) and quadruped support moving center of mass (referred to as 4S).
  • the durations of each of the eight motion stages are t1, t2, t3, t4, t5, t6, t7, and t8 shown in FIG. 6, for the convenience of description.
  • the eight movement stages are called the first movement stage, the second movement stage, and so on.
  • the control device 120 samples from each movement stage, and obtains the sampling moments 1 and 2 in the first movement stage as shown in FIG. 6 , the sampling moments 2, 3 and 4 in the second movement stage, and the third movement stage
  • the sampling instants in are 4, 5 and 6, the sampling instants in the fourth motion stage are 6, 7 and 8, and so on, to obtain multiple sampling instants.
  • the sampling moments belonging to the same shape in Fig. 6 represent the sampling moments belonging to the same movement stage, and the sampling moments of different shapes indicate that the two sampling moments belong to two different movement stages.
  • each motion stage is continuous in a sampling period
  • the end moment of a certain motion stage can be regarded as both the sampling moment in the motion stage and the sampling moment in the next motion stage, for example
  • the sampling time 6 in the above-mentioned FIG. 6 can be regarded as the sampling time in the third movement stage and the fourth movement stage at the same time.
  • S1.2 Determine a plurality of sampling moments from the movement duration, and determine the time interval between each sampling moment and the moment corresponding to the start point of the spatial path.
  • the control device 120 knows the starting time corresponding to the starting point of the spatial path and each sampling time, so the time interval between each sampling time and the starting time can be determined. Multiple time intervals can be obtained corresponding to multiple sampling moments.
  • relational expression between the position of the centroid at each sampling time and the contact force of the foot end is a pre-existing relational expression, and the relational expression is specifically as the formula (3) discussed above.
  • the control device 120 can determine the values of H k and w k in the above formula (14) according to the time interval corresponding to each sampling moment and the initial centroid position, so that only the centroid position variation coefficient and each sampling can be obtained. The relationship between the two unknown quantities of the foot contact force at the moment.
  • wk can take a fixed value, or the gravity of the footed robot 110, the initial position of the center of mass, and the value of the angular momentum L of the center of mass at the kth moment are substituted into the above formula , the value of w k at the k-th time is obtained.
  • the value of w k corresponding to each sampling moment is calculated.
  • the centroid angular momentum can be calculated from the body pose at each sampling moment.
  • the kth time is the selected sampling time, so the kth time can also be called the kth sampling time.
  • control device 120 obtains the stepping sequence of the footed robot 110, it is naturally possible to obtain the foot ends that need to rest in each motion stage, and the corresponding N i and N c corresponding to each sampling moment in the formula (14) can be obtained. value of .
  • the control device 120 can also obtain the world coordinate of the i-th contact point of the footed robot 110 in contact with the contact surface at each sampling moment in the case of determining the foot end that needs to land each time.
  • the position r ij under the system, or at least r ij can be represented according to the selected candidate footholds.
  • N i the contact positions of the four foot ends of the footed robot 110 and the contact surface have been determined, so N i can be determined, which is 1, N c is the number of landing points, 4, r
  • the value of ij is also known.
  • each sampling moment After determining the value of w k , the value of H k , the value of N i and N c at each sampling moment, the obtained values corresponding to each sampling moment are respectively substituted into the above formula (14), so as to obtain The change relationship between the coefficient of change of the center of mass position and the contact force of the foot end corresponding to each sampling time. If there are multiple sampling time, then each sampling time corresponds to a change relationship.
  • the constraint condition set includes spatial foothold constraints, and the meaning and specific expressions of the spatial foothold constraints can refer to the content discussed above, which will not be repeated here, and the spatial foothold constraints can refer to the above formula (19) for details.
  • the control device 120 can determine multiple sets of candidate results that satisfy the variation relationship and the set of constraints at each sampling moment.
  • Each set of candidate results includes a centroid position variation coefficient, and a target foothold corresponding to each of the multiple sampling moments.
  • the candidate result may also include the contact force f ijk corresponding to each sampling moment.
  • the control device 120 may determine the centroid position variation coefficient in any group of candidate results from the multiple groups of candidate results as the target centroid position variation coefficient, and determine the foothold in the candidate result as the target foothold.
  • control device 120 may determine the target centroid position variation coefficient and the target foothold from multiple sets of candidate results according to the target function.
  • the process of determining the candidate result by the control device 120 is described as an example as follows:
  • Step 1 For each sampling moment, according to the foot end in contact with the contact surface at each sampling moment, the body posture of the footed robot at each sampling moment, and the constraints of the space foothold, obtain the position change coefficient of the centroid and each sampling moment. The goal constraint relationship between the footholds of the moment.
  • the control device 120 can determine the number of the foot ends that land on the contact surface at each sampling moment, the rotation range of the joint corresponding to the foot end of the footed robot 110, and the length of the joint, so as to determine The values of the changes A ik , Ni, Bi ik and bi ik in the constraints of the space foothold are obtained, and then the target constraint relationship between the foothold at each sampling moment and the change coefficient c of the centroid position is obtained.
  • the control device 120 may, according to the number of feet that land on the contact surface at each sampling moment, the body posture of the footed robot 110 , the foot end of the footed robot 110 , The rotation range of the corresponding joint and the length of the joint, and then determine the values of the changes A ik , Ni, Bi ik and bi ik , so as to obtain the target constraint relationship between the target foothold and the polynomial coefficients at each sampling moment .
  • the rotation range of the joint corresponding to the foot end of the foot robot 110 and the length of the joint may be obtained by the control device 120 using the default joint length and the bending range of the joint, or the control device 120 obtains the joint length in advance.
  • the rotation ranges of the joints corresponding to any two foot ends of the footed robot 110 and the lengths of the joints may be the same, or may be different.
  • the meaning of the friction force constraint condition may refer to the content discussed above, which will not be repeated here.
  • the control device 120 can obtain the constraint relationship between the foot contact force and the foot end point corresponding to each sampling time according to the sampling time and the friction force constraint condition.
  • the meaning of the contact force constraint can refer to the content discussed above, which will not be repeated here.
  • the sampling time and the contact force constraint conditions are used to obtain the constraint relationship between the foothold and the contact force of the foot end corresponding to each sampling time.
  • Step 2 Combine the change relationship at each sampling time and the target constraint relationship at each sampling time to obtain multiple sets of candidate results.
  • the control device 120 may determine each group of candidate results satisfying these relationships according to the variation relationship of the sampling time and the target constraint relationship at each sampling time. For the meaning of the candidate result, reference may be made to the content discussed above, which will not be repeated here.
  • each group of determined candidate results should also satisfy these constraint conditions.
  • the candidate result should satisfy the following relation:
  • k represents any sampling moment, and the meanings of other letters in the above formula can refer to the content discussed above, and will not be repeated here.
  • v 0 is the centroid velocity and a 0 is the centroid acceleration.
  • control device 120 may randomly determine one candidate result from multiple groups of candidate results as the final target result.
  • Step 3 According to multiple groups of candidate results, minimize the objective function to obtain the objective result.
  • control device 120 combines the objective function to obtain the final objective result.
  • the control device 120 may determine the value of the objective function corresponding to each group of candidate results, and determine the candidate result corresponding to the smallest value of the objective function as the target result.
  • the objective function as the formula (26) discussed above as an example, the process of obtaining the value of the objective function corresponding to a set of candidate results by the control device 120 will be introduced:
  • the control device 120 obtains the foot contact force corresponding to each sampling moment according to the centroid position of each sampling moment in a set of candidate results, and determines the weighted sum of squares of the respective foot contact forces, thereby obtaining The value of J grf in the objective function shown in formula (26).
  • the weighted weight of each contact force may be the same, or may be different.
  • the control device 120 determines the weighted square of the difference between the centroid positions of every two adjacent sampling moments in the plurality of sampling moments, so as to obtain the value of J len in the objective function shown in formula (26).
  • the weighted weight for each centroid location may be the same, or it may be different.
  • the control device 120 determines the difference between the centroid position corresponding to the end point of the spatial path in the candidate result and the centroid position of the end point of the space path in the centroid state data, the centroid velocity and the centroid state corresponding to the end point of the spatial path in the candidate result
  • the weighted sum of squares of the difference between the centroid velocities at the end of the spatial path in the data, and the difference between the centroid acceleration at the end of the spatial path in the candidate results and the difference between the centroid acceleration at the end of the spatial path in the centroid state data so as to obtain the value of J tgt in the objective function shown in formula (26).
  • the control device 120 After obtaining the value of J grf , the value of J len and the value of J tgt in the objective function shown in formula (26), the control device 120 determines the value of J grf , the value of J len and the value of J tgt The sum of the values is used to determine the value of the objective function corresponding to the candidate result.
  • S505 Determine target motion control parameters according to the target centroid position variation coefficient and the target foothold.
  • control device 120 After the control device 120 obtains the target result, it also obtains the centroid position variation coefficient, and the control device 120 can obtain the centroid positions at multiple sampling times according to the centroid position variation coefficient and the corresponding time interval, and the specific calculation formula involved can refer to Equation (4) and Equation (8) above.
  • the control device 120 After the control device 120 obtains the centroid positions at multiple sampling moments, it can fit the centroid motion trajectory according to the centroid positions at the multiple sampling moments, or it can interpolate the centroid positions at the multiple sampling moments to obtain the footed robot 110 . trajectories of the center of mass. There are various fitting methods, which are not limited here. After obtaining the motion trajectory of the center of mass, it is equivalent to obtaining the position of the center of mass of the footed robot 110 at each moment.
  • FIG. 7 shows the motion trajectory of the center of mass fitted according to the target result determined in FIG. 6 . It can be seen from FIG. 7 that the position of the center of mass changes continuously with the movement stage.
  • control device 120 can calculate the whole body motion trajectory of the footed robot through inverse kinematics according to the position of the centroid of the footed robot 110 and the target landing point, and the whole body motion trajectory includes the joints of each joint of the footed robot 110 corresponding to each moment. angle.
  • the joint angle is used to represent the angle presented by the joint rotation of the footed robot. Then, through inverse dynamics and optimal control methods, the joint moments of multiple joints in the motion path of the robot are determined. Both joint torque and joint angle can be regarded as target motion control parameters.
  • control device 120 can realize the lifting or falling of each foot of the footed robot 110 by controlling each joint in each foot end of the footed robot 110, thereby driving the footed robot 110 to move along the moving path.
  • the control device 120 specifically controls the joint moments of multiple joints of the footed robot 110, so that at least one foot of the footed robot supports the footed robot to move, so that the true centroid position of the footed robot is kept at the above determined centroid position as much as possible. .
  • the determined position of the centroid satisfies the spatial landing point constraint, which improves the accuracy of determining
  • the obtained centroid position is consistent with the actual motion process of the footed robot, which improves the adaptability of the centroid position, and further improves the reasonableness of the joint torque determined based on the centroid position and the target foothold.
  • an embodiment of the present application provides a motion control device for a footed robot, which is equivalent to being set in the control device 120 discussed above.
  • the motion control device 800 for a footed robot includes:
  • an acquisition module 810 configured to acquire the centroid state data corresponding to the spatial path start point and the spatial path end point of the motion path;
  • the first determination module 820 is configured to determine the candidate footholds of each foot end in the motion path based on the start point of the space path and the end point of the space path;
  • the second determination module 830 is configured to determine the change relationship between the center of mass position variation coefficient and the foot contact force based on the center of mass state data;
  • the screening module 840 is configured to screen the target centroid position variation coefficient and the target foothold that satisfy the variation relationship under the constraint of the constraint condition set; wherein the constraint condition set includes the spatial foothold constraint condition;
  • the third determination module 850 is configured to determine the target motion control parameter according to the target center of mass position variation coefficient and the target foothold;
  • the control module 860 is configured to control the footed robot to move according to the movement path based on the target movement control parameter.
  • the centroid state data includes the initial centroid position of the starting point of the spatial path
  • the second determining module 830 is specifically configured to:
  • the second determining module 830 is specifically configured to:
  • the screening module 840 is specifically used to:
  • the screening module 840 is specifically used to:
  • each group of candidate results includes a center of mass position change coefficient and a foothold
  • the objective function is minimized, and the objective result is determined from the multiple groups of candidate results; wherein, the objective function is a quadratic term constructed according to the correlation of the candidate results.
  • the objective function includes at least one of the following:
  • the quadratic term of the difference between the end centroid acceleration at the end of the spatial path in the centroid state data and the centroid acceleration at the end of the spatial path in the candidate results.
  • the constraint condition set further includes at least one of the following:
  • the friction force constraint condition used to constrain the magnitude of the friction force between the foot end of the foot and the contact surface; wherein, the friction force between the foot end of the foot and the contact surface is determined according to the foot end contact force and the friction coefficient;
  • the contact force constraint used to constrain the component of the foot contact force in the normal direction to be less than or equal to the upper limit of the contact force.
  • the first determining module 820 is specifically configured to:
  • the candidate landing point corresponding to the foot end in contact with the contact surface is determined each time.
  • the third determining module 850 is specifically configured to:
  • the joint torque of the footed robot at any moment is determined.
  • motion control apparatus 800 for a footed robot shown in FIG. 8 can also be used to implement any of the motion control methods for a footed robot discussed above, which will not be repeated here.
  • an embodiment of the present application also provides a motion control device for a footed robot.
  • the footed robot motion control apparatus can be used to implement the functions of the footed robot 110 discussed above, or can be used to implement the functions of the above-mentioned control apparatus 120 .
  • a computer device 900 is represented in the form of a general-purpose computer device.
  • Components of computer device 900 may include, but are not limited to, at least one processor 910 , at least one memory 920 , and a bus 930 connecting various system components including processor 910 and memory 920 .
  • Bus 930 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a processor, or a local bus using any of a variety of bus structures.
  • Memory 920 may include readable media in the form of volatile memory, such as random access memory (RAM) 921 and/or cache memory 922 , and may further include read only memory (ROM) 923 .
  • the memory 920 may also include a program/utility 926 having a set (at least one) of program modules 925 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, which An implementation of a network environment may be included in each or some combination of the examples.
  • the processor 910 is configured to execute program instructions stored in the memory 920 and the like to implement the motion control method for the footed robot discussed above.
  • the computer device 900 may communicate with one or more external devices 940 (eg, keyboards, pointing devices, etc.), and may also communicate with one or more devices that enable the terminal to interact with the computer device 900, and/or communicate with the computer device 900 Any device (eg, router, modem, etc.) capable of communicating with one or more other devices. Such communication may take place through input/output (I/O) interface 950 . Also, the computer device 900 may communicate with one or more networks (eg, a local area network (LAN), a wide area network (WAN), and/or a public network such as the Internet) through a network adapter 960 . As shown, network adapter 960 communicates with other modules for computer device 900 via bus 930 . It should be understood that, although not shown, other hardware and/or software modules may be used in conjunction with computer device 900, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives and data backup storage systems.
  • the change relationship between the change coefficient of the center of mass position and the contact force of the foot end is determined, and the set of constraints and the change relationship are used to determine the foot robot.
  • the position change coefficient of the target centroid and the target foothold of the robot, and then the centroid position of the footed robot at each moment can be determined.
  • the footed robot foothold space constraints are considered.
  • centroid position of the robot is more consistent with the actual motion process of the footed robot, which improves the adaptability of the determined centroid position.
  • the movement trajectory of the center of mass of the footed robot can be automatically generated and the foothold of the footed robot can be automatically selected, which improves the intelligence level of the footed robot.
  • an embodiment of the present application provides a storage medium, where computer instructions are stored in the storage medium, and when the computer instructions are executed on the computer, the computer executes the motion control method for a footed robot discussed above.
  • an embodiment of the present application provides a computer program product, where the computer program product includes computer instructions, and the computer instructions are stored in a computer-readable storage medium.
  • the processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes any of the above-mentioned motion control methods for a footed robot.
  • the embodiments of the present application may be provided as a method, a system, or a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
  • computer-usable storage media including, but not limited to, disk storage, CD-ROM, optical storage, etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Combustion & Propulsion (AREA)
  • Mechanical Engineering (AREA)
  • Automation & Control Theory (AREA)
  • Transportation (AREA)
  • Chemical & Material Sciences (AREA)
  • Multimedia (AREA)
  • Electromagnetism (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Manufacturing & Machinery (AREA)
  • Human Computer Interaction (AREA)
  • Manipulator (AREA)

Abstract

一种足式机器人运动控制方法、装置、设备及介质，涉及机器人技术领域，控制方法包括：获取与运动路径的空间路径起点和空间路径终点对应的质心状态数据（S501）；基于空间路径起点和空间路径终点，确定在运动路径中各足端的候选落脚点（S502）；基于质心状态数据，确定质心位置变化系数与足端接触力的变化关系（S503）；在约束条件集合的约束下，筛选满足变化关系的目标质心位置变化系数和目标落脚点（S504），其中，约束条件集合包括空间落脚点约束条件；根据目标质心位置变化系数和目标落脚点，确定目标运动控制参数（S505）；基于目标运动控制参数，控制足式机器人按照运动路径进行运动（S506）。

Description

足式机器人运动控制方法、装置、设备及介质
本申请要求于2020年9月7日提交中国专利局、申请号为202010927483.5、发明名称为“足式机器人运动控制方法、装置、设备及介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及机器人技术领域，尤其涉及一种足式机器人运动控制方法、装置、设备及介质。
发明背景
随着计算机不断发展，逐渐出现了各种机器人，例如足式机器人，如何控制机器人的运动是机器人领域所关注的问题。在控制足式机器人运动之前，需要先确定出足式机器人质心运动轨迹，进而根据质心运动轨迹控制足式机器人。
相关技术中是根据足式机器人的运动方程，求解机器人的质心位置。但这种方式未考虑足式机器人的实际运动过程，使得确定出的质心位置与足式机器人的实际运动符合程度不高，导致足式机器人在该质心运动轨迹下很难完成运动过程，即相关技术中生成的质心位置适应性差。
发明内容
本申请实施例提供一种足式机器人运动控制方法、装置、设备及介质，用于提高生成的质心位置的适应性。
一方面，提供一种足式机器人运动控制方法，由足式机器人运动控制设备执行，包括：
获取与运动路径的空间路径起点和空间路径终点对应的质心状态数据；
基于所述空间路径起点和所述空间路径终点，确定在所述运动路径中各足端的候选落脚点；
基于所述质心状态数据，确定质心位置变化系数与足端接触力的变化关系；
在约束条件集合的约束下，筛选满足所述变化关系的目标质心位置变化系数和目标落脚点；其中，所述约束条件集合包括空间落脚点约束条件；
根据所述目标质心位置变化系数和所述目标落脚点，确定目标运动控制参数；
基于所述目标运动控制参数，控制所述足式机器人按照所述运动路径进行运动。
又一方面，提供一种足式机器人运动控制装置，包括：
获取模块，用于获取与运动路径的空间路径起点和空间路径终点对应的质心状态数据；
第一确定模块，用于基于所述空间路径起点和所述空间路径终点，确定在所述运动路径中各足端的候选落脚点；
第二确定模块，用于基于所述质心状态数据，确定质心位置变化系数与足端接触力的变化关系；
筛选模块，用于在约束条件集合的约束下，筛选满足所述变化关系的目标质心位置变化系数和目标落脚点；其中，所述约束条件集合包括空间落脚点约束条件；
第三确定模块，用于根据所述目标质心位置变化系数和所述目标落脚点，确定目标运动控制参数；
控制模块，用于基于所述目标运动控制参数，控制所述足式机器人按照所述运动路径进行运动。
本申请实施例提供一种足式机器人运动控制设备，包括：
存储器；
至少一个处理器，与所述存储器连接；
其中，所述存储器存储有可被所述至少一个处理器执行的指令，所述至少一个处理器通过执行所述存储器存储的指令实现如一方面中任一项所述的方法。
本申请实施例提供一种存储介质，所述存储介质存储有计算机指令，当所述计算机指令在计算机上运行时，使得计算机执行如一方面中任一项所述的方法。
附图简要说明
图1为本申请实施例提供的足式机器人运动控制方法的应用场景示例图；
图2为本申请实施例提供的控制设备的结构示例图；
图3为本申请实施例提供的工作空间的示例图；
图4为本申请实施例提供的一种足式机器人的结构示例图；
图5为本申请实施例提供的一种足式机器人运动控制方法的流程图；
图6为本申请实施例提供的一种采样时刻的分布示意图；
图7为本申请实施例提供的一种质心运动轨迹示意图；
图8为本申请实施例提供的一种足式机器人运动控制装置的结构示意图；
图9为本申请实施例提供的一种足式机器人运动控制设备的结构示意图。
实施方式
为了更好的理解本申请实施例提供的技术方案，下面将结合说明书附图以及具体的实施方式进行详细的说明。
为了便于本领域技术人员更好地理解本申请的技术方案，下面对本申请涉及的名词进行介绍。
机器人(Robot)：包括各类模拟人类行为或思想上模拟其他生物的机械(如机器狗，机器猫等)。在广义上某些电脑程序也被称为机器人。在当代工业中，機器人指能自动执行任务的人造机器人，用以取代或协助人类工作，可以是机电装置，或由电脑程序或电子电路控制。
足式机器人：泛指具有足端的机器人，足式机器人可以配置有一个或多个足端，例如两足、四足或六足等。每个足端对应配置有一个或多个关节。
运动路径：是指足式机器人需要完成的运动路径，运动路径的长度可以是任意的，具体不做限制。足式机器人完成运动路径所需的移动时长视为一个采样周期。移动时长中包括多个时刻，在实际确定质心位置时，从多个时刻中选择出的时刻称为采样时刻。
空间路径起点：又可以称为起始位置，是指机器人沿运动路径开始运动的位置，空间路径起点对应的时刻可以称为起始时刻。
空间路径终点：是指机器人沿运动路径停止运动的位置，空间路径终点对应的时刻可以称为终止时刻。
质心状态数据：是指用于描述机器人质心状态变化的数据，具体包括机器人的质心位置、质心速度或质心加速度中的一种或多种。质心位置：为机器人质量的中心位置，用于描述机器人的位置，机器人在不同的运动状态下，其质心位置会发生变化。质心速度可以由质心位置对时间间隔求一阶导得到，质心加速可以由质心位置对时间间隔求二阶导得到。为了便于描述，空间路径起点的质心位置可以称为起始质心位置，空间路径起点的质心速度可以称为起始质心速度，空间路径起点的质心加速度可以称为起始质心加速度。同理，空间路径终点的质心位置可以称为终点质心位置，空间路径终点的质心速度可以称为终点质心速度，空间路径终点的质心加速度可以称为终点质心加速度。
候选落脚点：是指在机器人的足端与接触面接触时，该足端与接触面可能的接触位置。例如，在某个时刻，机器人的候选落脚点包括A、B和C三个点。
质心位置变化系数：质心位置变化系数用于描述时间变化过程中质心位置变化参数。质心位置变化系数以矩阵形式表示，或者以向量形式表示等。质心位置变化系数与时间间隔能够共同表示特定时刻的质心位置，该时间间隔是指该特定时刻与空间路径起点对应时刻之间的时间差。
接触面：是指机器人的足端与环境所接触的面，接触面例如地面、或其它与足端接触的支撑物等。应当说明的是，由于路面不平等情况，足式机器人多个足端所对应的接触面可能是不同的。
质心运动轨迹：又可以称为质心位置运动轨迹，用于描述机器人在不同时刻的质心位置，实际是由包括机器人在不同时刻的质心位置组成。
接触点：又可以称为落脚的足端，是指机器人与接触面接触的足端，与接触面接触的足端的数量即为接触点的数量。当然，机器人在不同时刻，与接触面接触的足端并不完全相同。
迈步时序：用于描述机器人在完成运动路径过程中的步态情况，具体包括机器人沿运动路径运动过程中的至少一个运动阶段，以及每个运动阶段的持续时长。
足端接触力：是指机器人与接触面接触的足端与接触面之间的作用力大小。
约束条件集合：是指用于约束质心位置变化系数和/或落脚点的取值的关系式。约束条件集合包括一个或多个约束条件。本申请实施例中的约束条件集合至少包括空间落脚点约束条件，空间落脚点约束条件用于约束机器人的落脚点是位于机器人的工作空间内的，该空间落脚点约束条件具体是与机器人的质心位置变化系数。
目标运动控制参数：用于控制足式机器人运动所需的参数，具体包括足式机器人在任意时刻的关节角度，以及任意时刻的关节力矩等。
另外，本申请实施例中的“多个”是指两个或两个以上，“至少一个”是指一个或多个，“A和/或B”具体包括A、B以及A和B三种情况。
为了提高生成的质心位置的适应性，本申请实施例提供一种足式机器人运动控制方法，下面对本申请实施例涉及的足式机器人运动控制方法的设计思想进行介绍。
本申请实施例中先确定足式机器人足端对应的多个候选落脚点，根据足式机器人质心状态数据，确定质心位置变化系数与足端接触力的变化关系，并基于多个候选落脚点，利用该变化关系和约束条件集合，确定出足式机器人的目标质心位置变化系数和目标落脚点，进而根据目标质心位置变化系数确定足式机器人的质心运动轨迹，并根据质心运动轨迹和目标落脚点，确定目标运动控制参数，利用目标运动控制参数控制足式机器人沿运动路径运动。由于本申请实施例中将足式机器人的落脚点约束在工作空间内，使得确定出的质心位置以及目标落脚点均是满足空间落脚点约束条件，提高了确定出质心位置与实际运动过程的适应性。且，该方法实现了质心位置以及质心运动轨迹的自动生成，以及实现了自动选取落脚点，提高了控制足式机器人的自动化程度。
进一步地，本申请实施例中可以根据变化关系、约束条件集合，以及目标函数构建二次规划，利用求解二次规划的方式求解目标质心位置变化系数和目标落脚点。由于二次规划问题如果存在解，必然能够求解出全局最优解，将求解目标质心位置变化系数和目标落脚点转换为二次规划问题，使得能够求解出最优的目标质心位置变化系数和目标落脚点。
基于上述设计思想，下面对本申请实施例中足式机器人运动控制方法的应用场景进行介绍。
该足式机器人运动控制方法适用于控制各类足式机器人在各种环境下的各种步态。各类足式机器人例如双足机器人、四足机器人等。各种环境例如平地、不平地面、斜坡或楼梯等。各种步态例如双足行走、四足行走、四足小跑和随机步态等。
请参照图1，为一种足式机器人运动控制方法的应用场景图，或者可以理解为足式机器人运动控制系统的架构图，该架构图包括足式机器人110和控制设备120，下面对控制设备120与足式机器人110之间的交互进行示例介绍。
在一种可能的情况中，控制设备120与足式机器人110是相对独立的两个设备：
在该情况下，足式机器人110和控制设备120之间通过有线或无线进行通信，图1中是以足式机器人110和控制设备120之间利用通信网络实现通信为例。
在控制足式机器人110运动之前，控制设备120可以根据用户的操作，或者足式机器人110的任务，设置足式机器人110空间路径起点和空间路径终点的质心状态数据。或者，足式机器人110可以检测空间路径起点的质心状态数据，并将空间路径起点的质心状态数据上传至控制设备120。或者，控制设备120直接采集足式机器人110在空间路径起点的质心状态数据。
进一步地，控制设备120采集足式机器人110当前所处的环境图像，或接收足式机器人110上报的环境图像，控制设备120根据该环境图像确定足式机器人110在运动路径中需要落脚的 足端可能的候选落脚点。当然，控制设备120确定候选落脚点的方式有多种，下文进行具体介绍。
控制设备120根据质心状态数据，以及候选落脚点，确定足式机器人110的质心运动轨迹和落脚点，进而控制足式机器人110进行相应的运动。其中，确定质心运动轨迹和落脚点的内容将在下文中进行介绍。
控制设备120可以通过服务器或终端实现，服务器包括但不限于：独立的物理服务器、多个物理服务器构成的服务器集群、或者分布式系统，还可以是提供云服务、云数据库、云计算、云函数、云存储、网络服务、云通信、中间件服务、域名服务、安全服务、内容分发网络(Content Delivery Network，CDN)、以及大数据和人工智能平台等基础云计算服务的云服务器。终端例如手机、个人计算机、智能电视或便携式平板电脑等。
在另一种可能的情况中，控制设备120为足式机器人110的一部分：
控制设备120可以设置在足式机器人110的机体内，例如，控制设备120为足式机器人110中的内部处理器等。
在控制设备120控制足式机器人110运动之前，控制设备120可以从上位机接收运动指令，或根据用户的输入操作以获得运动指令，该运动指令可以指示足式机器人110执行某项任务，或者指示足式机器人110运动路径中的空间路径起点和空间路径终点。上位机可以为与控制设备120无线或有线连接的任意设备，上位机例如终端或服务器等。
同理，控制设备120采集足式机器人110质心状态数据。控制设备120可以根据足式机器人110当前所处的环境图像，确定足式机器人110每次落脚的足端可能的候选落脚点。控制设备120根据质心状态数据、以及候选落脚点，确定足式机器人的质心运动轨迹以及目标落脚点，以控制足式机器人110进行运动。其中，确定质心运动轨迹和落脚点的内容将在下文中进行介绍。
为了更清楚地介绍控制设备120的结构，下面结合图2所示的足式机器人运动控制系统进行示例介绍，图2中是以控制设备120包括视觉感知单元210、轨迹生成单元220和运动控制单元230为例。
其中，视觉感知单元210可以设置在足式机器人110上，例如搭载在足式机器人110的头部。视觉感知单元210例如摄像头、红外摄像头中的一种或多种。摄像头例如RGBD摄像头。
视觉感知单元210采集机器人状态。机器人状态包括足式机器人110空间路径起点和空间路径终点的质心状态数据。另外，视觉感知单元210还可以采集足式机器人110所处的环境图像，获得足式机器人110每次落脚可能的候选落脚点。还可以根据足式机器人110的运动路径等获得足式机器人110的迈步时序。
在获得质心状态数据、环境图像以及迈步时序之后，视觉感知单元210可以这些数据发送至轨迹生成单元220。
轨迹生成单元220接收根据机器人状态数据、迈步时序和候选落脚点，确定足式机器人110在多个时刻的质心位置和目标落脚点，并根据多个时刻的质心位置获得该足式机器人110的质心运动轨迹。进而根据质心运动轨迹和目标落脚点确定足式机器人110的全身运动轨迹，将全身运动轨迹和目标落脚点发送给运动控制单元230。
运动控制单元230可以根据全身运动轨迹和目标落脚点，确定该足式机器人110各个关节的关节力矩，按照各个关节力矩，控制足式机器人110的各个关节转动，从而实现足式机器人110的运动。
进一步地，运动控制单元230还可以监测足式机器人110移动的过程中实时的足式机器人状态，以保证足式机器人110能够稳定移动。
基于上述应用场景，下面对本申请实施例涉及的运动控制方法的总体思路进行介绍：
本申请实施例将足式机器人110在每个时刻的质心位置转换为质心位置变化系数、起始质心位置，以及该时刻与起始质心位置之间的时间间隔三者之间的关系。根据质心状态数据，将质心位置与足端接触力之间的关系转换为质心位置变化系数与足端接触力之间的变化关系，并基于该变化关系以及约束条件集合，进而求解出质心位置变化系数和目标落脚点，进而根据质心位置变化系数，获得足式机器人110的质心运动轨迹，并根据质心运动轨迹和目标落脚点确定足式机器人110的目标运动控制参数，利用目标运动控制参数控制足式机器人110的运动。
进一步地，在确定质心位置变化系数和目标落脚点时，可以结合约束条件集合和变化关系，将确定质心位置变化系数和目标落脚点的问题转换为一个混合整数二次规划问题，通过求解二次规划问题，以获得质心位置和目标落脚点。变化关系以及约束条件集合是控制设备120中预配置的，或者控制设备120从其他设备或网络资源获取的，或者由控制设备120创建的。下面对控制设备120创建变化关系和约束条件集合的方式进行示例介绍：
一，获得变化关系：
控制设备120可以从网络资源或其他设备获得每个时刻的质心位置与对应时刻的足端接触力之间的关系式，足端接触力的含义可以参照前文论述的内容，此处不再赘述。将该关系式中每个时刻的质心位置采用起始质心位置、质心位置变化系数和时间间隔共同表示，从而将每个时刻的质心位置与对应时刻的足端接触力之间的关系式转换为质心位置变化系数和足端接触力之间的变化关系。其中，变化关系用于表示质心位置变化系数和足端接触力之间的变化关系。在获得变化关系之后，控制设备120可以以任意形式存储该变化关系，任意形式例如函数形式、描述语句形式等。
下面对控制设备120创建变化关系的过程进行具体示例介绍：
1：获得足式机器人110的第一关系式。
第一关系式是指足式机器人110的质心动力学方程，用于表示足式机器人110的运动与所受外力之间的关系。第一关系式可以有多种表达形式，例如牛顿-欧拉方程，一种第一关系式示例如下：
Figure PCTCN2021114272-appb-000001
其中，m为足式机器人110的总质量，g∈R 3为重力加速度，p G∈R 3为足式机器人110的质心位置，
Figure PCTCN2021114272-appb-000002
为足式机器人110与接触面接触的第i个接触点的位置，或者也可以称为落脚点，即表示与接触面接触的足端的位置，L∈R 3为足式机器人110的质心角动量，
Figure PCTCN2021114272-appb-000003
表示由角动量对时间间隔求一阶导，f i∈R 3为第i个接触点的足端接触力，N c为接触点的个数，
Figure PCTCN2021114272-appb-000004
运算是指()的斜对角阵,
Figure PCTCN2021114272-appb-000005
表示由p G对时间间隔求二阶导。R 3表示坐标系下的三个坐标值。
应当说明的是，本申请实施例中在没有特定说明的情况下，各个量均是以世界坐标系下的表示结果。例如公式(1)中的各个变量均是在世界坐标系下的表示结果。
需要说明的是，公式(1)中的前三行是根据牛顿定律获得的，后三行是根据欧拉方程获得的。
进一步地，根据上述公式(1)中的前三行可知：
Figure PCTCN2021114272-appb-000006
将公式(2)代入公式(1)，获得如下公式：
Figure PCTCN2021114272-appb-000007
其中，
Figure PCTCN2021114272-appb-000008
G i∈R 3×3可以是3×3的矩阵。
2：将第一关系式中每个时刻的质心位置表示为从起始质心位置与经过时间间隔t后的质心位置变化量之和，获得第二关系式。
①：将质心位置设置表示为起始位置的质心位置到经过时间间隔t后的质心位置变化量之和，具体为：
p G＝p init+p t    (4)
其中，p init表示起始质心位置，p t表示经过时间间隔t后的质心变化量。
②：将公式(4)代入公式(3)，获得第二关系式如下：
Figure PCTCN2021114272-appb-000009
下面对上述第二关系式中的各项进行分析：
第一项
Figure PCTCN2021114272-appb-000010
与
Figure PCTCN2021114272-appb-000011
和p t呈线性关系，第二项
Figure PCTCN2021114272-appb-000012
为常数项，第三项
Figure PCTCN2021114272-appb-000013
存在
Figure PCTCN2021114272-appb-000014
第四项
Figure PCTCN2021114272-appb-000015
存在
Figure PCTCN2021114272-appb-000016
和
Figure PCTCN2021114272-appb-000017
的乘积。
作为一种实施例，当足式机器人110的身体姿态变化较小时，
Figure PCTCN2021114272-appb-000018
近似为0 3×1，或者，可以 根据足式机器人110预配置的每个时刻的身体姿态，确定出该第三项
Figure PCTCN2021114272-appb-000019
的取值。
3：将第二关系式中的第四项
Figure PCTCN2021114272-appb-000020
中的质心位置变化量设置为各个方向上的变化量，获得第三关系式：
Figure PCTCN2021114272-appb-000021
其中，p t＝p t xy+p t z,p t xy包括p t在x和y轴上的分量。
Figure PCTCN2021114272-appb-000022
是指质心位置变化量p t在xy轴构成的平面上的分量对时间间隔的二阶导数，也表示质心加速度在x、y轴构成的平面上的分量；
Figure PCTCN2021114272-appb-000023
是指质心位置变化量p t在z轴构成的平面上的分量对时间间隔的二阶导数，也表示质心加速度在z轴构成的平面上的分量；
Figure PCTCN2021114272-appb-000024
表示质心位置变化量p t在xy轴分量的斜对角阵，
Figure PCTCN2021114272-appb-000025
表示质心位置变化量p t在z轴分量的斜对角阵。
p t xy的z坐标为0，p t z包括p t在z轴上的分量，p t z对应的x和y的坐标为0。产生绕z轴的力矩为
Figure PCTCN2021114272-appb-000026
xy平面内某一方向上的力矩为
Figure PCTCN2021114272-appb-000027
由于
Figure PCTCN2021114272-appb-000028
和
Figure PCTCN2021114272-appb-000029
共线，因此
Figure PCTCN2021114272-appb-000030
另外，足式机器人110在z轴方向上的运动通常较为稳定，因此可以忽略
Figure PCTCN2021114272-appb-000031
以及
Figure PCTCN2021114272-appb-000032
此外，
Figure PCTCN2021114272-appb-000033
以及
Figure PCTCN2021114272-appb-000034
的绝对值较小，也可以忽略。在实际控制足式机器人110的过程中，也可以通过调整足端与接触面之间的足端作用力来获取上述公式(6)被忽略的项，也就是公式(6)中所涉及的力矩。
作为一种实施例，可以忽略第二关系式中的第五项，获得第四关系式：
Figure PCTCN2021114272-appb-000035
其中：
Figure PCTCN2021114272-appb-000036
其中，H 0和w为已知量，x t包括质心位置变化量p t和质心加速度
Figure PCTCN2021114272-appb-000037
为待定量。
4：将质心位置变化量设置为以时间为自变量的n阶多项式，具体如下：
Figure PCTCN2021114272-appb-000038
其中，T P＝[1t…t n]∈R 1×(n+1)，c *＝[c *,0c *,1…c *,n] T∈R n+1为多项式系数，t表示时间间隔，即该时刻与空间路径起点对应时刻之间的时间间隔，*代表x、y和z，c是指质心位置变化系数，其包含了所有多项式系数。在实际应用中在获得质心位置变化系数c后，根据上述公式(8)能计算出多个时刻中每个时刻的质心位置。
作为一种实施例，n的取值为大于或等于2的任意整数。
5：将上述公式(8)对时间间隔进行求二阶导，获得质心加速度，并根据质心加速度以及第四关系式，获得第五关系式。
其中，质心加速度具体表示如下：
Figure PCTCN2021114272-appb-000039
将公式(8)和公式(9)代入公式(7)获得如下的第五关系式为：
Figure PCTCN2021114272-appb-000040
其中，
Figure PCTCN2021114272-appb-000041
与时间间隔t相关。公式(10)表示了质心位置变化系数c与足端接触力f i之间的关系，该关系实际上是对质心位置与足端接触力之间的关系进行变换得到的。
作为一种实施例，在实际应用中，由于公式(10)中其它变量均是可以计算得到的，因此质心位置变化系数c和足端接触力f i之间呈线性关系。
6：在第五关系式中引入被选中的候选落脚点。被选中的候选落脚点是指从多个候选落脚点中被确定用于作为足端的落脚点，也可以视为目标落脚点。
足式机器人110中需要落脚的足端通常会有不止一个候选落脚点，即前文中涉及的r i∈{r ij|j＝1,2,3…N i},r ij表示第i足端的候选落脚点的位置，N i表示第i足端的候选落脚点的数量。
作为一种实施例，只有被选中的候选落脚点对应的足端接触力f i的取值域才不为零，因此可以引入二进制变量β ij∈{0,1}，j＝1,2,3…N i表示被选中的目标落脚点具体是哪一个候选落脚点，即：
Figure PCTCN2021114272-appb-000042
其中：
Figure PCTCN2021114272-appb-000043
公式(12)表示N i中二进制变量只有一个等于1，其余为0，表示只能选择一个候选落脚点作为目标落脚点，因此可以将被选中的候选落脚点引入上述第五关系式中，获得如下改写后的质心运动方程为：
Figure PCTCN2021114272-appb-000044
由于在不同时刻，f ij、H和w的取值可能不同，因此可以将上述公式(13)进一步表示为如下所示的变化关系：
Figure PCTCN2021114272-appb-000045
其中，k表示第k个时刻，f ijk表示第k个时刻且第i足端选中第j个候选落脚点时的足端接触力。H k表示第k个时刻所对应的H，H的具体表示可以参照前文。w k表示第k个时刻所对应的w，w的具体表示可以参照前文的内容。
应当说明的是，上述是对创建变化关系的过程进行示例说明，实际上述过程中可以利用其它的动力学方程描述足式机器人110的质心运动轨迹，进而对其它的动力学方程进行变换，以获得变化关系。
二：获得约束条件集合：
约束条件集合包括一个或多个约束条件，每个约束条件是用于约束质心位置或目标落脚点中的一个或两个的取值，每个约束条件的形式可以是不等式。约束条件集合至少包括空间落脚点约束条件，空间落脚点约束条件用于约束足式机器人110质心位置所对应的落脚点均是足式机器人的足端可达的。控制设备120可以是从网络资源、或其它设备获得约束条件集合、或者自行创建的约束条件集合，下面对控制设备120创建各个约束条件进行示例说明：
(1)获得空间落脚点约束条件：
1：将足式机器人110足端对应的工作空间近似为凸多面体，并获得该凸多面体中各个面的线性不等式的表示，联合各个面的线性不等式，获得凸多面体的线性不等式表示。凸多面体的线性不等式表示为第六关系式，该第六关系式具体表示如下：
S i Tx i≤d i     (15)
其中，S i＝[s il s i2…s il]∈R 3×l，d i＝[d il d i2…d il]∈R l，x i表示足式机器人110在该坐标系下的落脚点，d il是凸多面体中面与原点之间的距离，s il表示凸多面体中面对应的单位法向量，l表示凸多面体对应的面的数量，c表示质心位置变化系数，β ij表示目标落脚点，凸多面体是根据足式机器人110落脚的足端配置的关节的弯曲范围，以及关节的长度确定的。
具体的，控制设备120根据关节的弯曲范围和关节的长度，离散地确定出足式机器人110的足端相对于该关节可达到的位置。该关节是落脚的足端配置的关节，具体可以是与该足端直接连接的关节，或者与该足端通过关节所连接的其它关节。关节的弯曲范围是指关节能够弯曲的最小角度和最大角度组成的范围，例如关节的弯曲范围为0°～120°，该弯曲范围通常是已知的。关节的长度例如为1米。
在离散地确定出足式机器人110的足端相对于该关节可达到的多个位置后，控制设备120拟合多个位置，从而获得足式机器人的110的工作空间。而足式机器人110足端的工作空间一般呈非凸区域，但可以通过拟合方式将工作空间近似为凸多面体。凸多面体具体如图3中所示的凸多面体310，图3中所示的多个点中每个点表示足端相对于该关节可达到的位置。
在近似成凸多面体之后，控制设备120可以获得该凸多面体中每个面的线性不等式表示，每个面的线性不等式表示具体表示为s il Tx i≤d il。控制设备120联合该凸多面体各个面的不等式，从而获得如上述的公式(15)。
在具体实施时，上述公式(15)可以在关节的局部固定坐标系中进行，关节的局部固定坐标系是指以该关节的局部为坐标原点建立的坐标系，即关节的局部固定坐标系的坐标原点与世界坐标系下的坐标原点可能不同。关节可以是与落脚的足端相关的任一关节。
例如，请参照图4，为足式机器人110的结构示意图，该足式机器人110包括多个关节410，4个足端420，每个足端420配置有多个关节，关节的局部固定坐标系如图4中所示的坐标系。
进一步地，在该局部固定坐标系下，各个变量均可以根据该关节的局部固定坐标系分解为多个变量。
例如，图4中的足端接触力f i可以沿局部固定坐标系进行分解为n i、f i1和f i2，质心位置p G也可以沿局部固定坐标系进行分解。
如前文论述的内容，控制设备120已提前确定出了需要落脚的足端所对应的候选落脚点430，控制设备120可以根据上述公式(15)判断选择的某个候选落脚点430是否位于足端的凸多面体内。例如，可以判断足式机器人110在时间间隔t后，第i足端的落脚点是否位于该凸多面体中。
2：将落脚点位置r i转换为局部固定坐标系的表示，获得第七关系式。
由于上述公式(15)为局部固定坐标系下的表示结果，因此需要将世界坐标系(又可以称为全局坐标系)下的落脚的足端的位置r i转换为局部固定坐标系下的表示结果x i，具体转换过程表示为：
Figure PCTCN2021114272-appb-000046
其中，R 0∈R 3×3为旋转矩阵，表示足式机器人110在当前时刻的身体姿态，p il∈R 3为足式机器人110中第i足端的关节相对于足式机器人110质心在身体坐标系中的位置，R il∈R 3为关节在局部固定坐标系中相对于身体坐标系的身体姿态。p il和R il均为常量。
作为一种实施例，当足式机器人110的身体姿态变化较小时，R 0可以为常数，或者控制设备120可以根据当前时刻足式机器人110的身体姿态确定。
3：将第七关系式中质心位置表示为与时间相关的n阶多项式，并代入第六关系式中，获得第八关系式：
控制设备120联合上述公式(4)、(8)、(16)和(15)，获得第八关系式为：
A ic+B ir i≤b i   (17)
其中：
Figure PCTCN2021114272-appb-000047
Figure PCTCN2021114272-appb-000048
Figure PCTCN2021114272-appb-000049
4：将选中的落脚点和时间引入第八关系式，获得空间落脚点约束条件：
将被选中的候选落脚点引入第八关系式，具体结合公式(11)和公式(17)，获得第九关系式表示为：
Figure PCTCN2021114272-appb-000050
由于不同时刻，该空间落脚点约束条件中对应的某些量取值不同，因此可以将时间引入第九关系式，获得空间落脚点约束条件为：
Figure PCTCN2021114272-appb-000051
其中，A ik、B ik和b ik分别表示第k个时刻对应的A i、B i和b i。
作为一种实施例，除了空间落脚点约束条件，约束条件集合还可以包括用于约束接触面接触的足端与接触面之间摩擦力大小的摩擦力约束条件、或用于约束足端接触力在法线方向上的接触力小于或等于接触力上限的接触力约束条件中的一种或两种。其中，落脚的足端与接触面之间的摩擦力根据足端接触力和摩擦系数确定。
应当说明的是，由于足式机器人110在每个时刻的足端接触力大小是不同的，因此每个时刻，足端与接触面之间的摩擦力也是不同的，因此摩擦力约束条件实际上约束了每个时刻接触面接触的足端与接触面之间的摩擦力大小，而接触力约束条件是约束了每个时刻足端接触力在法线方向上的接触力大小。
下面对控制设备120创建摩擦力约束条件和接触力约束条件的方式进行示例：
(2)获得摩擦力约束条件：
1：确定每个接触力对应的摩擦力约束表示，获得第十关系：
每个足端接触力f i(i＝1,2…N c)均受到摩擦力约束，具体可以近似为如下第十关系式：
Figure PCTCN2021114272-appb-000052
其中，N i＝-[μ in i-o i μ in i+o i μ in i-t i μ in i+t i]∈R 3×4，n i表示第i足端的落脚点的法向量，o i表示第i足端的落脚点的在一个切线方向上的向量，t i第i足端的落脚点位置的另一个切线方向上的向量，μ i表示足端与接触面之间的摩擦系数。
应当说明的是，足端所接触的接触面材质不同时，对应的摩擦系数的取值也可能不同，在同一个时刻，不同足端所接触的接触面不同，不同足端与接触面之间的摩擦力系数也可能不同。
2：在十关系式中引入时间和落脚的足端，获得摩擦力约束条件。
具体的，足端对应的位置r ij处的法向量为n ij，足端接触力为f ij，获得第十一关系式：
Figure PCTCN2021114272-appb-000053
其中，N ij表示第i足端被选中的第j个候选落脚点时，所对应的N i的取值 。
在不同时刻，足端接触力大小不同，因此可以在公式(17)中引入时刻，获得摩擦力约束条件为：
Figure PCTCN2021114272-appb-000054
其中，f ijk表示第k个时刻且第i足端选中第j个候选落脚点时的足端接触力。
(3)获得接触力约束条件：
足式机器人110在运动过程中，足端与接触面之间的足端接触力如果过大，容易损害足式机器人110的部件，因此可以设置接触力约束条件，约束足式机器人110与接触面之间的足端接触力大小，避免足式机器人110每次运动与接触面之间的作用力过大。下面对控制设备120创建接触力约束条件的方式进行示例：
1：将足端接触力在法线方向上的分力设置为小于或等于接触力上限，获得第十二关系式：
Figure PCTCN2021114272-appb-000055
其中，
Figure PCTCN2021114272-appb-000056
表示接触力上限，其取值可以根据实际情况设置，例如可以参照足式机器人的重力设置。
2：在第十二关系式中引入时间和落脚的足端，获得第三约束条件：
每次落脚的足端与接触面之间均存在足端接触力，因此公式(23)可以改写为：
Figure PCTCN2021114272-appb-000057
由于每个时刻的足端接触力不同，在公式(19)中引入时间，具体表示为：
Figure PCTCN2021114272-appb-000058
其中，
Figure PCTCN2021114272-appb-000059
表示第i足端选中第j个候选落脚点时所对应的
Figure PCTCN2021114272-appb-000060
的取值，f ijk表示第k个时刻且第i足端选中第j个候选落脚点时的足端接触力。
由于变化关系、约束条件集合所对应的关系式较少，而需要求解的未知量较多，因此满足变化关系、约束条件集合的质心位置变化系数和落脚点可能不止一个，因此在实际求解上述中的质心位置变化系数和目标落脚点时，可以从满足变化关系、约束条件集合的取值中随机确定出目标质心位置变化系数和目标落脚点。或者，为了获得较佳的质心位置变化系数和落脚点，本申请实施例中还可以引入目标函数，利用目标函数进一步地筛选出最优的质心位置变化系数和落脚点，进而将最优的质心位置变化系数确定为目标质心位置变化系数，将最优的落脚点确定为目标落脚点。
由于求解二次规划问题必然地能够获得相应的解，因此本申请实施例中可以将确定目标质心位置变化系数以及目标落脚点转化为一个二次规划问题，为此，本申请实施例中的目标函数至少包括一个或多个变量的二次项，一个或多个变量可以是满足变化关系和约束条件集合的候选结果相关的任意的变量。二次项可以根据变量的二次方构建。
在一种实施例中，目标函数包括如下的A1～A5中的至少一种：
A1：在运动路径中足端接触力的变化量相关的二次项；
A2：在运动路径中质心位置的变化量的二次项；
A3：质心状态数据中所述空间路径终点的终止质心位置与候选结果中在空间路径终点的质心位置的差值的二次项；
A4：质心状态数据中所述空间路径终点的终止质心速度与候选结果中在空间路径终点的质心速度的差值的二次项；
A5：质心状态数据中所述空间路径终点的终止质心加速度与候选结果中在空间路径终点的 质心加速度的差值的二次项。
下面对上述各个二次项的作用进行分析：
A1的作用：足端接触力的变化量可以用于优化足端与接触面之间的作用力的分配，使得足式机器人110在行走过程中足端与接触面之间的作用力分配更为均匀。
A2的作用：质心位置变化量反应了质心运动轨迹的长度，有利于降低质心运动轨迹的震荡幅度。
A3～A5的作用：有利于降低计算结果与实际结果之间的误差。
联合上述A1～A5，一种目标函数的表达式如下：
J grf+J len+J tgt    (26)
其中，J grf为在运动路径中所有足端接触力的加权平方和，J len为每相邻两个时刻之间的质心位置变化量之间的差值的加权平方和，J tgt为质心状态数据中空间路径终点的终止质心位置与候选结果中在空间路径终点的质心位置的差值、候选结果中的终止位置对应的质心速度与终止位置的状态数据对应的质心速度之间的差值、以及候选结果中的终止位置对应的质心加速度与终止位置的状态数据对应的质心加速度之间的差值三者的加权平方和。
在介绍完构建变化关系、约束条件集合和目标函数之后，下面结合图5所示的足式机器人运动控制方法的流程，对本申请实施例具体如何利用构建出的变化关系和约束条件集合，确定质心位置以及目标落脚点的过程进行示例介绍，请参照图5，该方法可由图9所示的足式机器人运动控制设备执行，包括：
S501，获取与运动路径的空间路径起点和空间路径终点对应的质心状态数据。
在控制足式机器人110运动时，控制设备120是通过控制足式机器人110的足端的抬起和落下，进而实现足式机器人110的移动，使得足式机器人110能够完成从空间路径起点至空间路径终点的整个运动路径的运动。空间路径起点为足式机器人110当前时刻所在的位置。空间路径终点为足式机器人110要达到的位置，空间路径终点可以是提前设定的，或者控制设备120根据足式机器人110所需执行的任务确定的。
质心状态数据的含义以及获取方式可以参照前文论述的内容，此处不再赘述。质心状态数据可以包括质心位置。或者，控制设备120可以根据质心位置、质心速度或质心加速度中的一种，求解出另外两种的取值。质心位置、质心速度或质心加速度均可以通过坐标系中的坐标表示或通过向量表示等。
作为一种实施例，运动路径、空间路径起点和空间路径终点是与选择的足式机器人110的采样周期相关的，运动路径、空间路径起点和空间路径终点三者可以根据实际需要进行灵活设定。
例如，控制设备120可以将足式机器人110当前时刻对应的位置确定为空间路径起点，控制设备120可以将第3s所对应的位置确定为空间路径终点，在下一个采样周期时，可以选择第3s作为足式机器人110运动的空间路径起点。
S502，基于空间路径起点和所述空间路径终点，确定在运动路径中各足端的候选落脚点。
足式机器人110从空间路径起点到空间路径终点过程中，可能有一个或多个足端落脚，每个足端可能落脚一次或多次，具体与设定的空间路径起点和所述空间路径终点相关，控制设备120可以预先为每次需要落脚的足端确定多个候选落脚点。候选落脚点是指足式机器人110在运动路径中可能的落脚点。落脚点可以用世界坐标系中的坐标表示或向量表示等。
在具体实现时，控制设备120可以通过视觉感知单元210采集足式机器人110的环境图像，并构建环境图像中各像素点与世界坐标系之间的转换关系。控制设备120通过环境图像，以及转换关系，确定沿空间路径起点到空间路径终点的移动方向上可能的候选落脚点。
控制设备120具体可以根据环境图像，识别从空间路径起点到空间路径终点中的障碍物，根据转换关系，确定沿空间路径起点到空间路径终点所对应的移动方向上的非障碍物的位置，并将确定出的位置作为候选落脚点。
或者，控制设备120通过视觉感知单元210采集环境的三维点云图，根据环境的三维点云图，从三维点云图中确定候选落脚点，具体可以从三维点云图中确定非障碍物的位置作为候选落脚点。其中，视觉感知单元210为RGBD摄像头时，可以采集得到三维点云图，或者例如通过采集足式机器人120当前所处的多个环境图像，对多个环境图像进行三维重建，获得三维点 云图。
或者，控制设备120可以从空间路径起点到空间路径终点的移动方向上随机选择多个位置作为候选落脚点。
由于足式机器人110在运动路径过程可能不止一次落脚，控制设备110可以确定出机器人在运动路径中的所有可能的候选落脚点，该所有可能的候选落脚点为每次落脚所对应候选落脚点。或者，在确定候选落脚点时，可以根据足式机器人110的运动速度，确定足式机器人110可能落脚的大区域，并根据上述任一方式依次从每次落脚对应的大区域中筛选出候选落脚点。
例如，请继续参照图4，足式机器人110的当前位置如图4所示，控制设备120分别确定足端的候选落脚点430包括图4中所示的地面上的多个圆圈。
S503，基于质心状态数据，确定质心位置变化系数与足端接触力的变化关系。
其中，变化关系例如前文论述的公式(14)，变化关系的含义可以参照前文论述的内容，此处不再赘述。
如果要实现对足式机器人110的运动控制，理论上需要获得足式机器人110从空间路径起点到空间路径终点中任意时刻的质心位置，但这样的计算量较大，为此，本申请实施例中控制设备120可以通过获得运动路径中多个时刻中每个时刻的质心位置，基于这多个时刻的质心位置进而确定出质心运动轨迹。
参照前文公式(14)所示的变化关系，该变化关系中除了质心位置变化系数和足端接触力，还包括H k和w k等参数，因此可以通过质心状态数据等已知量确定这些参数的取值，进而获得只包含质心位置变化系数与足端接触力两个未知量的变化关系。
下面对控制设备120获得变化关系的方式进行具体示例介绍：
S1.1，获得多个采样时刻。
控制设备120可以根据运动路径的长度以及足式机器人110的运动速度，获得足式机器人完成运动路径所需的移动时长，或者控制设备120直接预配置有足式机器人110完成运动路径所需的移动时长。
在获得移动时长之后，控制设备120可以从移动时长中随机采样，获得多个采样时刻。随机采样以获得采样时刻的方式较为简单。
或者，控制设备120根据足式机器人110的迈步时序，从每个运动阶段的持续时长中分别进行采样，以获得多个采样时刻。由于每个运动阶段都有其对应的采样时刻，这样可以保证各个运动阶段都能对应存在采样时刻，利于提高后期确定出的质心运动轨迹的准确性。
作为一种实施例，任意相邻两个采样时刻之间的时间间隔可以是相同的，也可以是不相同的，不相同是指任意相邻两个采样时刻之间的时间间隔是不完全相同的，或者存在相邻两个采样时刻之间的时间间隔不相同。
作为一种实施例，采样时刻的个数越多，采样时刻的分布越合理，确定出的质心运动轨迹可靠性越高。但采样时刻越多，后续求解质心位置变化系数以及目标落脚点所构建的关系式的数量越多，求解质心位置变化系数以及目标落脚点所需时间越长，因此合理规划采样点的数量极为重要，在本申请实施例中，采样时刻至少包括每个运动阶段的起始时刻和终止时刻，以及每个运动阶段中的至少一中间时刻。中间时刻是指该运动阶段的起始时刻和终止时刻之间的任意时刻。
例如，将足式机器人110的四足行走步态设置为一个采样周期，控制设备120将该采样周期中足式机器人110的运动过程依次划分为八个运动阶段，八个运动阶段具体为：四足支撑移动质心(简称为4S)、迈右后足(简称为HR)、迈右前足(简称为FR)、四腿支撑移动质心(简称为4S)、四足支撑移动质心(简称为4S)、迈左后足(简称为HL)、迈左前足(简称为FL)和四足支撑移动质心(简称为4S)。
请参照图6所示的采样时刻分布示意图，八个运动阶段每个运动阶段的持续时长分别为图6中所示的t1、t2、t3、t4、t5、t6、t7和t8，为便于描述，八个运动阶段分别称为第一运动阶段、第二运动阶段，以此类推。控制设备120从每个运动阶段进行采样，获得如图6所示的第一运动阶段中的采样时刻为1和2，第二运动阶段中的采样时刻为2、3和4，第三运动阶段中的采样时刻为4、5和6，第四运动阶段中的采样时刻为6、7和8，依次类推，获得多个采样时刻。其中，图6中属于同一种形状的采样时刻表示属于同一个运动阶段中的采样时刻，不同形状的 采样时刻表示两个采样时刻属于两个不同的运动阶段。
应当说明的是，由于一个采样周期内各个运动阶段是连续的，因此某个运动阶段的结束时刻既可以视为该运动阶段中的采样时刻，又可以视为下一个运动阶段的采样时刻，例如上述图6中的采样时刻6可以同时视为第三运动阶段和第四运动阶段中的采样时刻。
S1.2，从移动时长中确定出多个采样时刻，并确定每个采样时刻与空间路径起点对应的时刻之间的时间间隔。
控制设备120已知空间路径起点对应的起始时刻，以及每个采样时刻，因此可以确定出每个采样时刻与起始时刻之间的时间间隔。多个采样时刻对应可以获得多个时间间隔。
S1.3，针对每个采样时刻，根据每个采样时刻对应的时间间隔、起始质心位置、以及每个采样时刻的质心位置与足端接触力之间的关系式，获得质心位置变化系数与每个采样时刻的足端接触力之间的变化关系。
其中，每个采样时刻的质心位置与足端接触力之间的关系式为预存的关系式，该关系式具体如前文论述的公式(3)。
控制设备120根据每个采样时刻对应的时间间隔，以及起始质心位置，可以确定上述公式(14)中的H k和w k的取值，从而可以获得只存在质心位置变化系数与每个采样时刻的足端接触力两个未知量的变化关系。
如前文论述的内容，
Figure PCTCN2021114272-appb-000061
将每个采样时刻对应的时间间隔代入两个公式中，可以计算出每个时刻对应的H k的取值。
w k可以取固定值，或者将足式机器人110的重力，以及起始质心位置，以及第k个时刻的质心角动量L的取值代入上述的公式
Figure PCTCN2021114272-appb-000062
中，获得第k个时刻的w k的取值。以此类推，从而计算出每个采样时刻对应的w k的取值。质心角动量可以通过每个采样时刻的身体姿态计算得到。第k个时刻是被选择的采样时刻，因此第k个时刻又可以称为第k个采样时刻。
进一步地，由于控制设备120获得足式机器人110的迈步时序，自然也就能获得各个运动阶段需要落脚的足端，可以获得公式(14)中对应的每个采样时刻对应的N i和N c的取值。另外，在迈步时序确定的情况下，控制设备120在确定每次需要落脚的足端的情况下，也能对应获得每个采样时刻足式机器人110与接触面接触的第i个接触点在世界坐标系下的位置r ij，或至少能根据被选中的候选落脚点表示出r ij。
请继续参照图6所示的例子，下面对各个运动阶段涉及的变量的取值进行分析：
(1)在首个4S阶段，足式机器人110的四个足端与接触面的接触位置已确定，因此可以确定N i，取值为1，N c为落脚点的个数为4，r ij的取值也已知。
(2)当足式机器人110处在HR阶段中的采样点时，由于右后足端在空中尚未触地，其对应的N i为0，即无需考虑对右后足端的任何约束，N c为落脚点的个数和r ij的取值依旧和(1)中相同。
(3)当足式机器人110处在紧接着的FR及随后各阶段中某时刻时，右后足端已触地，其位置取决于落脚点的选取，故对应右后足端的N i设不为1，r ij为右后腿候选落脚点位置，由于右前足端尚在迈腿过程中，并未触地，其对应N i设为0，以此类推。
在确定每个采样时刻的w k的取值、H k的取值、N i和N c的取值之后，将获得的每个采样时刻对应的值分别代入上述公式(14)中，从而获得质心位置变化系数与每个采样时刻所对应的足端接触力之间的变化关系，如果有多个采样时刻，那么每个采样时刻都对应一个变化关系。
S504，在约束条件集合的约束下，筛选满足变化关系的目标质心位置变化系数和目标落脚点。
其中，约束条件集合包括空间落脚点约束条件，空间落脚点约束条件的含义及具体表达式可以参照前文论述的内容，此处不再赘述，空间落脚点约束条件具体可参照前文公式(19)。
由于根据每个采样时刻的变化关系，以及约束条件集合并不能获得唯一解，因此，控制设备120可以确定出满足每个采样时刻的变化关系和约束条件集合的多组候选结果。每一组候选结果包括质心位置变化系数，以及多个采样时刻中每个采样时刻对应的目标落脚点。候选结果还可以包括每个采样时刻对应的接触力f ijk。控制设备120可以从多组候选结果中将任一组候选结果中的质心位置变化系数确定为目标质心位置变化系数，并将该候选结果中的落脚点确定为 目标落脚点。
或者，控制设备120可以根据目标函数，从多组候选结果中确定出目标质心位置变化系数和目标落脚点。下面对控制设备120确定候选结果的过程进行示例介绍：
步骤1：针对每个采样时刻，根据每个采样时刻与接触面接触的足端、足式机器人在每个采样时刻的身体姿态、以及空间落脚点约束条件，获得质心位置变化系数与每个采样时刻的落脚点之间的目标约束关系。
空间落脚点约束条件的表达式可以参照前文论述的公式(19)，在公式(19)中除了质心位置变化系数c和被选中的候选落脚点β ij之外，还包括一些需要控制设备120求解的变化量，具体包括公式(19)中的A ik、B ik、Ni、b ik，下面对控制设备120确定这些变化量的方式进行示例说明。
当足式机器人110的身体姿态变化较小时，控制设备120可以根据每个采样时刻与接触面落脚的足端的数量、足式机器人110足端对应的关节的转动范围、以及关节的长度，从而确定出空间落脚点约束条件中的变化量A ik、Ni、B ik和b ik的取值,进而获得每个采样时刻的落脚点与质心位置变化系数c的目标约束关系。
或者，当足式机器人110的身体姿态发生变化较大不可忽略时，控制设备120可以根据每个采样时刻与接触面落脚的足端的数量、足式机器人110的身体姿态、足式机器人110足端对应的关节的转动范围、以及关节的长度，进而确定出变化量A ik、Ni、B ik和b ik的取值,从而获得每个采样时刻的目标落脚点和多项式系数之间的目标约束关系。
作为一种实施例，足式机器人110足端对应的关节的转动范围、以及关节的长度可以是控制设备120采用默认的关节的长度以及关节的弯曲范围获得的，或者控制设备120预先获得的该足式机器人110的关节的长度以及关节的弯曲范围。其中，足式机器人110的任意两个足端所对应的关节的转动范围、以及关节的长度可以是相同的，或者也可以是不同的。
在一种实施例中，当约束条件集合还包括摩擦力约束条件时，其中，摩擦力约束条件的含义可以参照前文论述的内容，此处不再赘述。控制设备120可以根据采样时刻，以及摩擦力约束条件，获得每个采样时刻所对应的落脚点与足端接触力之间的约束关系。
在一种实施例中，当约束条件集合还包括接触力约束条件时，接触力约束条件的含义可以参照前文论述的内容，此处不再赘述，控制设备120可以根据接触力上限的取值、采样时刻、以及接触力约束条件，获得每个采样时刻所对应的落脚点与足端接触力之间的约束关系。
例如，继续参照图6所示的例子，
(1)对于首个4S阶段中的采样点，由于初始四只脚与地面接触位置均已确定，N i取值为1，β ij的取值为1，r ij为足端的初始位置。
(2)当机器人处在HR阶段中的采样点时，由于右后腿在空中尚未触地，其对应的N i取值为1，即无需考虑对右后足端的任何约束，其余不变。
(3)当机器人处在紧接着的FR及随后各阶段中某时刻时，右后腿已触地，其位置取决于落脚点的选取，故对应右后足端的N i取值不为1，相应的β ij设为二进制变量，r ij为右后足端候选落脚点，由于右前足端尚在迈腿过程中、未触地，其对应的N i取值为0，以此类推。
步骤2：联合每个采样时刻的变化关系，每个采样时刻的目标约束关系，获得多组候选结果。
控制设备120可以根据采样时刻的变化关系，每个采样时刻的目标约束关系，确定满足这些关系的各组候选结果。候选结果的含义可以参照前文论述的内容，此处不再赘述。
作为一种实施例，如果约束条件集合还包括摩擦力约束条件和/或接触力约束条件时，则确定出的每组候选结果还应当满足这些约束条件。
例如，候选结果具体应当满足如下关系式：
Figure PCTCN2021114272-appb-000063
其中，k表示任意一个采样时刻，上述公式中其它字母的含义可以参照前文论述的内容，此处不再赘述。v 0表示质心速度，a 0表示质心加速度。
在一种实施例中，控制设备120可以从多组候选结果中随机确定出一个候选结果，作为最终的目标结果。
步骤3：根据多组候选结果，最小化目标函数，获得目标结果。
在本申请实施例中，控制设备120结合目标函数，以获得最终的目标结果。
具体的，控制设备120在获得候选结果之后，可以确定每组候选结果所对应的目标函数的取值，将目标函数的取值最小所对应的候选结果确定为目标结果。下面以目标函数为前文论述的公式(26)为例，对控制设备120获得一组候选结果对应的目标函数的取值的过程进行介绍：
(1)控制设备120根据一组候选结果中多个采样时刻中每个采样时刻的质心位置，获得每个采样时刻对应的足端接触力，确定各个足端接触力的加权平方和，从而获得公式(26)所示的目标函数中的J grf的取值。其中，每个接触力的加权权重可以是相同的，或者可以是不同的。
(2)控制设备120确定多个采样时刻中每相邻两个采样时刻的质心位置的差值的加权平方，，从而获得公式(26)所示的目标函数中J len的取值。每个质心位置的加权权重可以是相同的，或者可以是不同的。
(3)控制设备120确定候选结果中在空间路径终点对应的质心位置与质心状态数据中在空间路径终点的质心位置之间的差值、候选结果中在空间路径终点对应的质心速度与质心状态数据中在空间路径终点的质心速度之间的差值、以及候选结果中在空间路径终点对应的质心加速度与质心状态数据中在空间路径终点的质心加速度之间的差值三者的加权平方和，从而获得公式(26)所示的目标函数中的J tgt的取值。
在获得公式(26)所示的目标函数中J grf的取值、J len的取值和J tgt的取值之后，控制设备120确定J grf的取值、J len的取值和J tgt的取值之和，从而确定该候选结果对应的目标函数的取值。
S505，根据所述目标质心位置变化系数和所述目标落脚点，确定目标运动控制参数。
控制设备120在获得目标结果后，也就获得了质心位置变化系数，控制设备120可以根据质心位置变化系数以及对应的时间间隔，获得多个采样时刻的质心位置，其中涉及的具体计算公式可以参照前文中的公式(4)和公式(8)。
控制设备120在获得多个采样时刻的质心位置之后，可以根据多个采样时刻的质心位置拟合出质心运动轨迹，或者，可以对多个采样时刻的质心位置进行插值，从而获得足式机器人110的质心运动轨迹。拟合方式有多种，此处不作限定。在获得质心运动轨迹之后，也就相当于获得了足式机器人110在每个时刻的质心位置。
例如，请参照图7，表示根据图6中确定出的目标结果拟合出的质心运动轨迹，从图7可以看出随着运动阶段的不同，质心位置不断发生变化。
进一步地，控制设备120可以根据足式机器人110的质心位置以及目标落脚点，通过逆运动学计算出足式机器人全身运动轨迹，全身运动轨迹包括足式机器人110在各个时刻所对应的各个关节的关节角度。该关节角度用于表示足式机器人的关节转动后所呈现的角度。然后，通过逆动力学和优化控制方法，确定机器人在运动路径中多个关节的关节力矩。关节力矩以及关节角度均可以视为目标运动控制参数。
S506，基于目标运动控制参数，控制足式机器人按照运动路径进行运动。
具体的，控制设备120可以通过控制足式机器人110各个足端中的各个关节，来实现足式 机器人的各个足的抬起或落下，从而带动足式机器人110沿着移动路径移动。控制设备120具体通过控制足式机器人110的多个关节的关节力矩，使足式机器人的至少一个足支撑足式机器人移动，使足式机器人的真实质心位置尽量地保持在上述确定出的质心位置。
本申请实施例中，由于在确定足式机器人110的质心位置以及目标落脚点时，引入了空间落脚点约束条件，因此使得确定出的质心位置是满足该空间落脚点约束条件的，提高了确定出的质心位置与足式机器人实际运动过程的相符合程度，提升了质心位置的适应性，进而提高了基于质心位置和目标落脚点确定的关节力矩的合理程度。
基于同一发明构思，本申请实施例提供一种足式机器人运动控制装置，该装置相当于设置在前文论述的控制设备120中，请参照图8，该足式机器人运动控制装置800包括：
获取模块810，用于获取与运动路径的空间路径起点和空间路径终点对应的质心状态数据；
第一确定模块820，用于基于空间路径起点和空间路径终点，确定在运动路径中各足端的候选落脚点；
第二确定模块830，用于基于质心状态数据，确定质心位置变化系数与足端接触力的变化关系；
筛选模块840，用于在约束条件集合的约束下，筛选满足变化关系的目标质心位置变化系数和目标落脚点；其中，约束条件集合包括空间落脚点约束条件；
第三确定模块850，用于根据目标质心位置变化系数和目标落脚点，确定目标运动控制参数；
控制模块860，用于基于目标运动控制参数，控制足式机器人按照运动路径进行运动。
在一种实施例中，质心状态数据包括空间路径起点的起始质心位置，第二确定模块830具体用于：
获得足式机器人完成运动路径所需的移动时长；
从移动时长中确定出多个采样时刻，并确定每个采样时刻与空间路径起点对应的时刻之间的时间间隔；
针对每个采样时刻，根据每个采样时刻对应的时间间隔、起始质心位置、以及每个采样时刻的质心位置与足端接触力之间的关系式，获得质心位置变化系数与每个采样时刻的足端接触力之间的变化关系。
在一种实施例中，第二确定模块830具体用于：
根据每个采样时刻的身体姿态，确定每个采样时刻的质心角动量；
根据每个采样时刻的质心角动量、每个采样时刻对应的时间间隔、起始质心位置、以及每个采样时刻的质心位置与足端接触力之间的关系式，获得质心位置变化系数与每个采样时刻的足端接触力之间的变化关系。
在一种实施例中，筛选模块840具体用于：
针对每个采样时刻，根据每个采样时刻与接触面接触的足端、足式机器人在每个采样时刻的身体姿态、以及空间落脚点约束条件，确定质心位置变化系数与每个采样时刻的落脚点之间的目标约束关系；
筛选出满足每个采样时刻对应的目标约束关系，以及每个采样时刻的足端接触力与质心位置变化系数之间变化关系的目标质心位置变化系数和目标落脚点。
在一种实施例中，筛选模块840具体用于：
在约束条件集合的约束下，获得满足变化关系的多组候选结果；其中，每组候选结果包括质心位置变化系数和落脚点；
根据多组候选结果，最小化目标函数，从多组候选结果中确定出目标结果；其中，目标函数是根据候选结果的相关量构建的二次项。
在一种实施例中，目标函数包括如下至少一种：
在运动路径中足端接触力的变化量相关的二次项；
在运动路径中质心位置的变化量的二次项；
质心状态数据中空间路径终点的终止质心位置与候选结果中在空间路径终点的质心位置的差值的二次项；
质心状态数据中空间路径终点的终止质心速度与候选结果中在空间路径终点的质心速度的差值的二次项；
质心状态数据中空间路径终点的终止质心加速度与候选结果中在空间路径终点的质心加速度的差值的二次项。
在一种实施例中，约束条件集合还包括如下至少一种：
用于约束落脚的足端与接触面之间的摩擦力大小的摩擦力约束条件；其中，落脚的足端与接触面之间的摩擦力根据足端接触力和摩擦系数确定；
用于约束足端接触力在法线方向上的分量小于或等于接触力上限的接触力约束条件。
在一种实施例中，第一确定模块820具体用于：
采集足式机器人当前所处环境的三维点云图；
根据足式机器人在运动路径中每次与接触面接触的足端，从三维点云图中，确定出每次与接触面接触的足端所对应的候选落脚点。
在一种实施例中，第三确定模块850具体用于：
根据目标质心位置变化系数，确定足式机器人在任意时刻的质心位置；
根据足式机器人在任意时刻的质心位置，以及目标落脚点，确定足式机器人在任意时刻的关节力矩。
应当说明的是，图8中所示的足式机器人运动控制装置800还可以用于实现前文论述的任一的足式机器人运动控制方法，此处不再赘述。
基于同一发明构思，本申请实施例还提供了一种足式机器人运动控制设备。该足式机器人运动控制设备能够用于实现前文论述的足式机器人110的功能，或能够用于实现前文控制设备120的功能。
请参照图9，计算机设备900以通用计算机设备的形式表现。计算机设备900的组件可以包括但不限于：至少一个处理器910、至少一个存储器920、连接不同系统组件(包括处理器910和存储器920)的总线930。
总线930表示几类总线结构中的一种或多种，包括存储器总线或者存储器控制器、外围总线、处理器或者使用多种总线结构中的任意总线结构的局域总线。
存储器920可以包括易失性存储器形式的可读介质，例如随机存取存储器(RAM)921和/或高速缓存存储器922，还可以进一步包括只读存储器(ROM)923。存储器920还可以包括具有一组(至少一个)程序模块925的程序/实用工具926，这样的程序模块925包括但不限于：操作系统、一个或者多个应用程序、其它程序模块以及程序数据，这些示例中的每一个或某种组合中可能包括网络环境的实现。处理器910用于执行存储器920存储的程序指令等实现前文论述的足式机器人运动控制方法。
计算机设备900可以与一个或多个外部设备940(例如键盘、指向设备等)通信，还可与一个或者多个使得终端能与计算机设备900交互的设备通信，和/或与使得该计算机设备900能与一个或多个其它设备进行通信的任何设备(例如路由器、调制解调器等)通信。这种通信可以通过输入/输出(I/O)接口950进行。并且，计算机设备900还可以通过网络适配器960与一个或者多个网络(例如局域网(LAN)，广域网(WAN)和/或公共网络，例如因特网)通信。如图所示，网络适配器960通过总线930与用于计算机设备900的其它模块通信。应当理解，尽管图中未示出，可以结合计算机设备900使用其它硬件和/或软件模块，包括但不限于：微代码、设备驱动器、冗余处理器、外部磁盘驱动阵列、RAID系统、磁带驱动器以及数据备份存储系统等。
本申请实施例中，根据足式机器人在空间路径起点和空间路径终点的质心状态数据，确定质心位置变化系数与足端接触力的变化关系，并利用约束条件集合以及该变化关系，确定足式机器人的目标质心位置变化系数和目标落脚点，进而可以确定出足式机器人在每个时刻的质心位置，由于在求解目标质心位置变化系数和目标落脚点时，考虑了足式机器人落脚点空间约束条件，使得基于目标质心位置变化系数确定的质心位置和目标落脚点均是满足落脚点空间约束条件的，即确定的质心位置以及目标落脚点均是足式机器人实际可达的，从而使得确定的足式机器人的质心位置与足式机器人的实际运动过程相符程度更高，提高了确定的质心位置的适应性。且，本申请实施例中能够实现自动生成足式机器人的质心运动轨迹和实现自动选取足式机器人的落脚点，提高了足式机器人智能化程度。
基于同一发明构思，本申请实施例提供一种存储介质，所述存储介质存储有计算机指令， 当所述计算机指令在计算机上运行时，使得计算机执行前文论述的足式机器人运动控制方法。
基于同一发明构思，本申请实施例提供一种计算机程序产品，该计算机程序产品包括计算机指令，该计算机指令存储在计算机可读存储介质中。计算机设备的处理器从计算机可读存储介质读取该计算机指令，处理器执行该计算机指令，使得该计算机设备执行上述的任一足式机器人运动控制方法。
本领域内的技术人员应明白，本申请的实施例可提供为方法、系统、或计算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
显然，本领域的技术人员可以对本申请进行各种改动和变型而不脱离本申请的精神和范围。这样，倘若本申请的这些修改和变型属于本申请权利要求及其等同技术的范围之内，则本申请也意图包含这些改动和变型在内。

Claims (20)

  1. 一种足式机器人运动控制方法，由足式机器人运动控制设备执行，其特征在于，包括：
    获取与运动路径的空间路径起点和空间路径终点对应的质心状态数据；
    基于所述空间路径起点和所述空间路径终点，确定在所述运动路径中各足端的候选落脚点；
    基于所述质心状态数据，确定质心位置变化系数与足端接触力的变化关系；
    在约束条件集合的约束下，筛选满足所述变化关系的目标质心位置变化系数和目标落脚点；其中，所述约束条件集合包括空间落脚点约束条件；
    根据所述目标质心位置变化系数和所述目标落脚点，确定目标运动控制参数；
    基于所述目标运动控制参数，控制所述足式机器人按照所述运动路径进行运动。
  2. 根据权利要求1所述的方法，其特征在于，所述质心状态数据包括空间路径起点的起始质心位置；基于所述质心状态数据，确定质心位置变化系数与足端接触力的变化关系，包括：
    获得所述足式机器人完成所述运动路径所需的移动时长；
    从所述移动时长中确定出多个采样时刻，并确定每个采样时刻与所述空间路径起点对应的时刻之间的时间间隔；
    针对每个采样时刻，根据每个采样时刻对应的时间间隔、所述起始质心位置、以及每个采样时刻的质心位置与足端接触力之间的关系式，获得所述质心位置变化系数与每个采样时刻的足端接触力之间的变化关系。
  3. 根据权利要求2所述的方法，其特征在于，针对每个采样时刻，根据每个采样时刻对应的时间间隔、所述起始质心位置、以及每个采样时刻的质心位置与足端接触力之间的关系式，获得所述质心位置变化系数与每个采样时刻的足端接触力之间的变化关系，包括：
    根据每个采样时刻的身体姿态，确定每个采样时刻的质心角动量；
    根据每个采样时刻的质心角动量、每个采样时刻对应的时间间隔、所述起始质心位置、以及每个采样时刻的质心位置与足端接触力之间的关系式，获得质心位置变化系数与每个采样时刻的足端接触力之间的变化关系。
  4. 根据权利要求2所述的方法，其特征在于，在约束条件集合的约束下，筛选满足所述变化关系的目标质心位置变化系数和目标落脚点，包括：
    针对每个采样时刻，根据每个采样时刻与接触面接触的足端、所述足式机器人在每个采样时刻的身体姿态、以及空间落脚点约束条件，确定质心位置变化系数与每个采样时刻的落脚点之间的目标约束关系；
    筛选出满足所述每个采样时刻对应的目标约束关系，以及所述每个采样时刻的足端接触力与质心位置变化系数之间变化关系的目标质心位置变化系数和目标落脚点。
  5. 根据权利要求1～4任一项所述的方法，其特征在于，在约束条件集合的约束下，筛选满足所述变化关系的目标质心位置变化系数和目标落脚点，包括：
    在约束条件集合的约束下，获得满足所述变化关系的多组候选结果；其中，每组候选结果包括质心位置变化系数和落脚点；
    根据所述多组候选结果，最小化目标函数，从所述多组候选结果中确定出目标结果；其中，所述目标函数是根据候选结果的相关量构建的二次项。
  6. 根据权利要求5所述的方法，其特征在于，所述目标函数包括如下至少一种：
    在所述运动路径中足端接触力的变化量相关的二次项；
    在所述运动路径中质心位置的变化量的二次项；
    所述质心状态数据中所述空间路径终点的终止质心位置与候选结果中在空间路径终点的质心位置的差值的二次项；
    所述质心状态数据中所述空间路径终点的终止质心速度与候选结果中在空间路径终点的质心速度的差值的二次项；
    所述质心状态数据中所述空间路径终点的终止质心加速度与候选结果中在空间路径终点的质心加速度的差值的二次项。
  7. 根据权利要求1～4任一项所述的方法，其特征在于，所述约束条件集合还包括如下至少 一种：
    用于约束落脚的足端与接触面之间的摩擦力大小的摩擦力约束条件；其中，落脚的足端与接触面之间的摩擦力根据足端接触力和摩擦系数确定；
    用于约束足端接触力在法线方向上的分量小于或等于接触力上限的接触力约束条件。
  8. 根据权利要求1～4任一项所述的方法，其特征在于，基于所述空间路径起点和所述空间路径终点，确定在所述运动路径中各足端的候选落脚点，包括：
    采集所述足式机器人当前所处环境的三维点云图；
    根据所述足式机器人在所述运动路径中每次与接触面接触的足端，从所述三维点云图中，确定出所述每次与接触面接触的足端所对应的候选落脚点。
  9. 根据权利要求1～4任一项所述的方法，其特征在于，根据所述目标质心位置变化系数和所述目标落脚点，确定目标运动控制参数，包括：
    根据所述目标质心位置变化系数，确定所述足式机器人在任意时刻的质心位置；
    根据所述足式机器人在任意时刻的质心位置，以及所述目标落脚点，确定所述足式机器人在任意时刻的关节力矩。
  10. 一种足式机器人运动控制装置，其特征在于，包括：
    获取模块，用于获取与运动路径的空间路径起点和空间路径终点对应的质心状态数据；
    第一确定模块，用于基于所述空间路径起点和所述空间路径终点，确定在所述运动路径中各足端的候选落脚点；
    第二确定模块，用于基于所述质心状态数据，确定质心位置变化系数与足端接触力的变化关系；
    筛选模块，用于在约束条件集合的约束下，筛选满足所述变化关系的目标质心位置变化系数和目标落脚点；其中，所述约束条件集合包括空间落脚点约束条件；
    第三确定模块，用于根据所述目标质心位置变化系数和所述目标落脚点，确定目标运动控制参数；
    控制模块，用于基于所述目标运动控制参数，控制所述足式机器人按照所述运动路径进行运动。
  11. 根据权10所述的装置，其特征在于，所述质心状态数据包括所述空间路径起点的起始质心位置，所述第二确定模块进一步用于：
    获得所述足式机器人完成所述运动路径所需的移动时长；
    从所述移动时长中确定出多个采样时刻，并确定每个采样时刻与所述空间路径起点对应的时刻之间的时间间隔；
    针对每个采样时刻，根据每个采样时刻对应的时间间隔、所述起始质心位置、以及每个采样时刻的质心位置与所述足端接触力之间的关系式，获得所述质心位置变化系数与每个采样时刻的足端接触力之间的变化关系。
  12. 根据权11所述的装置，其特征在于，所述第二确定模块进一步用于：
    根据每个采样时刻的身体姿态，确定每个采样时刻的质心角动量；
    根据每个采样时刻的质心角动量、每个采样时刻对应的时间间隔、所述起始质心位置、以及每个采样时刻的质心位置与足端接触力之间的关系式，获得质心位置变化系数与每个采样时刻的足端接触力之间的变化关系。
  13. 根据权11所述的装置，其特征在于，所述筛选模块进一步用于：
    针对每个采样时刻，根据每个采样时刻与接触面接触的足端、所述足式机器人在每个采样时刻的身体姿态、以及空间落脚点约束条件，确定质心位置变化系数与每个采样时刻的落脚点之间的目标约束关系；
    筛选出满足所述每个采样时刻对应的目标约束关系，以及所述每个采样时刻的足端接触力与质心位置变化系数之间变化关系的目标质心位置变化系数和目标落脚点。
  14. 根据权10～13任一项所述的装置，其特征在于，所述筛选模块进一步用于：
    在约束条件集合的约束下，获得满足所述变化关系的多组候选结果；其中，每组候选结果包括质心位置变化系数和落脚点；
    根据所述多组候选结果，最小化目标函数，从所述多组候选结果中确定出目标结果；其中， 所述目标函数是根据候选结果的相关量构建的二次项。
  15. 根据权14所述的装置，其特征在于，所述目标函数包括如下至少一种：
    在所述运动路径中足端接触力的变化量相关的二次项；
    在所述运动路径中质心位置的变化量的二次项；
    所述质心状态数据中所述空间路径终点的终止质心位置与候选结果中在空间路径终点的质心位置的差值的二次项；
    所述质心状态数据中所述空间路径终点的终止质心速度与候选结果中在空间路径终点的质心速度的差值的二次项；
    所述质心状态数据中所述空间路径终点的终止质心加速度与候选结果中在空间路径终点的质心加速度的差值的二次项。
  16. 根据权10-13任一项所述的装置，其特征在于，所述约束条件集合还包括如下至少一种：
    用于约束落脚的足端与接触面之间的摩擦力大小的摩擦力约束条件；其中，落脚的足端与接触面之间的摩擦力根据足端接触力和摩擦系数确定；
    用于约束足端接触力在法线方向上的分量小于或等于接触力上限的接触力约束条件。
  17. 根据权10-13任一项所述的装置，其特征在于，所述第一确定模块进一步用于：
    采集所述足式机器人当前所处环境的三维点云图；
    根据所述足式机器人在所述运动路径中每次与接触面接触的足端，从所述三维点云图中，确定出所述每次与接触面接触的足端所对应的候选落脚点。
  18. 根据权10-13任一项所述的装置，其特征在于，所述第三确定模块进一步用于：
    根据所述目标质心位置变化系数，确定所述足式机器人在任意时刻的质心位置；
    根据所述足式机器人在任意时刻的质心位置，以及所述目标落脚点，确定所述足式机器人在任意时刻的关节力矩。
  19. 一种足式机器人运动控制设备，其特征在于，包括：
    存储器；
    至少一个处理器，与所述存储器连接；
    其中，所述存储器存储有可被所述至少一个处理器执行的指令，所述至少一个处理器通过执行所述存储器存储的指令实现如权利要求1～9中任一项所述的方法。
  20. 一种存储介质，其特征在于，所述存储介质存储有计算机指令，当所述计算机指令在计算机上运行时，使得计算机执行如权利要求1～9中任一项所述的方法。
PCT/CN2021/114272 2020-09-07 2021-08-24 足式机器人运动控制方法、装置、设备及介质 WO2022048472A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP21863539.9A EP4116782A4 (en) 2020-09-07 2021-08-24 METHOD, APPARATUS AND DEVICE FOR CONTROLLING LEGGED ROBOT MOVEMENT, AND MEDIUM
US17/945,659 US20230016514A1 (en) 2020-09-07 2022-09-15 Legged robot motion control method, apparatus, and device, and storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010927483.5A CN112051797B (zh) 2020-09-07 2020-09-07 足式机器人运动控制方法、装置、设备及介质
CN202010927483.5 2020-09-07

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/945,659 Continuation US20230016514A1 (en) 2020-09-07 2022-09-15 Legged robot motion control method, apparatus, and device, and storage medium

Publications (1)

Publication Number Publication Date
WO2022048472A1 true WO2022048472A1 (zh) 2022-03-10

Family

ID=73607764

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/114272 WO2022048472A1 (zh) 2020-09-07 2021-08-24 足式机器人运动控制方法、装置、设备及介质

Country Status (4)

Country Link
US (1) US20230016514A1 (zh)
EP (1) EP4116782A4 (zh)
CN (1) CN112051797B (zh)
WO (1) WO2022048472A1 (zh)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112051797B (zh) * 2020-09-07 2023-12-26 腾讯科技（深圳）有限公司 足式机器人运动控制方法、装置、设备及介质
CN112114589B (zh) * 2020-09-07 2024-02-09 腾讯科技（深圳）有限公司 足式机器人运动控制方法、装置、设备及介质
CN114474034B (zh) * 2020-10-26 2024-05-28 腾讯科技（深圳）有限公司 足式机器人运动控制方法、装置、设备及介质
CN112731941B (zh) * 2020-12-29 2024-03-05 深圳市优必选科技股份有限公司 双足机器人路径规划方法、装置和双足机器人
CN112783175B (zh) * 2021-01-04 2022-09-16 腾讯科技（深圳）有限公司 质心轨迹确定方法、装置、足式机器人、设备及介质
US11833680B2 (en) * 2021-06-25 2023-12-05 Boston Dynamics, Inc. Robot movement and online trajectory optimization
CN117940258A (zh) * 2021-09-15 2024-04-26 索尼集团公司 机器人装置和机器人控制方法

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140257559A1 (en) * 2011-10-11 2014-09-11 Commissariat A L'energie Atomique Et Aux Energies Alternatives Method of Controlling a Robot
CN106826813A (zh) * 2016-12-12 2017-06-13 杭州宇芯机器人科技有限公司 基于曲线拟合建模和多重约束落足点评估的机器人稳定运动控制方法
CN107450555A (zh) * 2017-08-30 2017-12-08 唐开强 一种基于深度强化学习的六足机器人实时步态规划方法
CN107562052A (zh) * 2017-08-30 2018-01-09 唐开强 一种基于深度强化学习的六足机器人步态规划方法
CN108237532A (zh) * 2016-12-23 2018-07-03 深圳光启合众科技有限公司 多足机器人的步态控制方法、装置和机器人
CN108333931A (zh) * 2018-01-25 2018-07-27 北京理工大学 一种面向崎岖地形的四足机器人双层结构步态规划方法
US20200016750A1 (en) * 2015-09-15 2020-01-16 Boston Dynamics, Inc. Determination of Robotic Step Path
CN111483532A (zh) * 2020-06-02 2020-08-04 山东大学 一种静步态上楼梯的四足机器人运动控制方法
CN111506094A (zh) * 2020-04-14 2020-08-07 腾讯科技（深圳）有限公司 质心位置确定方法、装置、足式机器人及存储介质
CN111605642A (zh) * 2020-06-11 2020-09-01 哈尔滨工业大学 六足机器人的自由容错步态规划方法、装置及存储介质
CN112051797A (zh) * 2020-09-07 2020-12-08 腾讯科技（深圳）有限公司 足式机器人运动控制方法、装置、设备及介质

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102011106321A1 (de) * 2011-07-01 2013-01-03 Kuka Laboratories Gmbh Verfahren und Steuermittel zum Steuern eines Roboters
CN102591344B (zh) * 2012-03-05 2013-12-11 中国人民解放军国防科学技术大学 四足仿生机器人的时位控制方法
CN103217903B (zh) * 2013-04-07 2016-01-20 南京邮电大学 基于双平衡控制机制的仿人足球机器人全向踢球方法
US10059392B1 (en) * 2016-06-27 2018-08-28 Boston Dynamics, Inc. Control of robotic devices with non-constant body pitch
CN108519773B (zh) * 2018-03-07 2020-01-14 西安交通大学 一种结构化环境下无人驾驶车辆的路径规划方法
CN110209048A (zh) * 2019-05-20 2019-09-06 华南理工大学 基于动力学模型的机器人时间最优轨迹规划方法、设备
CN111267991B (zh) * 2020-01-19 2021-05-11 之江实验室 一种双足机器人下肢结构尺寸的设计方法
CN111208826B (zh) * 2020-02-18 2023-06-16 杭州云深处科技有限公司 四足机器人溜蹄步态规划方法、装置、设备及可读介质

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140257559A1 (en) * 2011-10-11 2014-09-11 Commissariat A L'energie Atomique Et Aux Energies Alternatives Method of Controlling a Robot
US20200016750A1 (en) * 2015-09-15 2020-01-16 Boston Dynamics, Inc. Determination of Robotic Step Path
CN106826813A (zh) * 2016-12-12 2017-06-13 杭州宇芯机器人科技有限公司 基于曲线拟合建模和多重约束落足点评估的机器人稳定运动控制方法
CN108237532A (zh) * 2016-12-23 2018-07-03 深圳光启合众科技有限公司 多足机器人的步态控制方法、装置和机器人
CN107450555A (zh) * 2017-08-30 2017-12-08 唐开强 一种基于深度强化学习的六足机器人实时步态规划方法
CN107562052A (zh) * 2017-08-30 2018-01-09 唐开强 一种基于深度强化学习的六足机器人步态规划方法
CN108333931A (zh) * 2018-01-25 2018-07-27 北京理工大学 一种面向崎岖地形的四足机器人双层结构步态规划方法
CN111506094A (zh) * 2020-04-14 2020-08-07 腾讯科技（深圳）有限公司 质心位置确定方法、装置、足式机器人及存储介质
CN111483532A (zh) * 2020-06-02 2020-08-04 山东大学 一种静步态上楼梯的四足机器人运动控制方法
CN111605642A (zh) * 2020-06-11 2020-09-01 哈尔滨工业大学 六足机器人的自由容错步态规划方法、装置及存储介质
CN112051797A (zh) * 2020-09-07 2020-12-08 腾讯科技（深圳）有限公司 足式机器人运动控制方法、装置、设备及介质

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4116782A4

Also Published As

Publication number Publication date
EP4116782A4 (en) 2023-10-25
CN112051797A (zh) 2020-12-08
CN112051797B (zh) 2023-12-26
EP4116782A1 (en) 2023-01-11
US20230016514A1 (en) 2023-01-19

Similar Documents

Publication Publication Date Title
WO2022048472A1 (zh) 足式机器人运动控制方法、装置、设备及介质
WO2022048266A1 (zh) 足式机器人运动控制方法、装置、设备及介质
WO2022089195A1 (zh) 足式机器人运动控制方法、装置、设备、介质及程序
JP7324932B2 (ja) 動的計画コントローラ
CN111506094A (zh) 质心位置确定方法、装置、足式机器人及存储介质
CN113190029B (zh) 基于深度强化学习的四足机器人适应性步态自主生成方法
WO2022089247A1 (zh) 足式机器人运动控制方法、装置、设备、介质及程序
Pap et al. Optimization of the hexapod robot walking by genetic algorithm
CN114723807A (zh) 质心轨迹确定方法、装置、足式机器人、设备及介质
WO2021208917A1 (zh) 质心位置确定方法、装置、足式机器人及存储介质
WO2024021749A1 (zh) 足式机器人的运动控制方法、装置、足式机器人、存储介质及程序产品
US11787045B2 (en) Robot with an inverse kinematics (IK)-based controller for retargeting input motions
WO2024051408A1 (zh) 移动轨迹确定方法、装置、计算机设备及存储介质
WO2024021767A1 (zh) 控制足式机器人的方法、装置、设备、足式机器人、计算机可读存储介质和计算机程序产品
US20230415340A1 (en) Artificial intelligence-actuated robot
Wang et al. Adaptive System for Object Detection and Picking Based on Efficient Convolutional Neural Networks
Mănoiu-Olaru et al. Hexapod robot. Virtual models for preliminary studies
CN114253134A (zh) 双足机器人的驱动方法、装置、控制器及介质
Mănoiu-Olaru et al. Basic walking simulations and gravitational stability analysis for a hexapod robot using Matlab
Lopes Quadruped manipulator for potential agricultural applications
Liu et al. Dynamic modeling and torque distribution of a climbing hexapod robot
CN114683281A (zh) 足式机器人运动控制方法、装置、电子设备及存储介质
CN116952508A (zh) 接触碰撞测试方法、接触碰撞测试系统
CN117754599A (zh) 拖拽示教的实现方法及装置、存储介质、电子设备
CN116985110A (zh) 控制足式机器人的方法和装置及足式机器人

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21863539

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2021863539

Country of ref document: EP

Effective date: 20221005

NENP Non-entry into the national phase

Ref country code: DE