JPWO2007139135A1 - Robot and control device - Google Patents

Abstract

A robot and control device that optimally distributes the force required for movement to any number of contact points in the space and generates torque for each connecting part. The drive control device 3 included in the robot of the present invention receives the input of the target motion information (step S1), and derives the target ZMP and the target acting force according to the target motion indicated by the received target motion information. (Step S2), the force applied to each contact portion is derived so that the derived applied force is optimally distributed based on the norm minimum norm (Step S3), and each connection is determined from the applied force distributed to the force applied to each contact portion. A torque value is calculated as each control target value of each actuator that drives the unit (step S4), and the calculated torque value is output as a control target value to each actuator that drives the connecting unit (step S5). Each actuator receives a torque value as a control target value and operates based on the received torque value.

Description

  The present invention relates to a robot including a plurality of drivable connecting portions, and a control device that controls the operation of the robot, and more particularly, to a robot and a control device that realize stable motion.

Research and development of robots such as biped robots that can imitate human walking forms are being promoted by various companies and research institutions, and various control methods for maintaining a stable posture during exercise Has been proposed.

  For example, in Patent Document 1, when a robot joint is driven so as to follow a dynamic model, and there is a difference between the dynamic model and the actual robot posture, an ankle joint is used to adjust the floor reaction force of the dynamic model. In addition, by driving the hip joint, the posture is stabilized by absorbing the difference between the dynamic model and the actual robot.

  In such a biped robot, the floor reaction force received from the floor by the robot is detected by a floor reaction force sensor, and the target value of the angular velocity or angle of the leg joint is calculated based on the detected floor reaction force. Various motion control is performed by controlling the actuator that drives the joint based on the calculated target value. The joint angular velocity and angle target values are calculated based on the principle of the inverted pendulum by determining the target trajectories of the robot waist and torso parts so that the ZMP (zero moment point) of the robot follows the target ZMP. Do to derive. In order to derive the target trajectories of the waist and torso of the robot, the target trajectories of each joint, that is, the target values of angular velocity and angle, are calculated by inverse kinematic calculation from the target trajectories of the waist and torso.

Further, there is the following Non-Patent Document 1 as a technology related to robots.
JP-A-5-337849 Jaeheung Park, Osama Katie, “A Whole Body Control Framework for Humanoid Operating in Human Environment” Proceeding of ITriple International Conference on Robotics Automation, USA, Itripe International Conference on Robotics Automation (Jaeheung Park, Oussama Khatib, “A whole-body control framework for humanoids operating in human environments”, Proceedings of IEEE International Conference on Robotics and Automation, United States of America, IEEE International Conference on Robotics and Automation, May 15, 2006, p. (1963-1969)

  However, as described above, in the conventional robot that controls the actuator with the angular velocity or angle as the target value, the floor reaction force detected by the floor reaction force sensor is fed back, and the angular velocity or angle of the joint is adjusted to the target value to adjust the floor reaction force. Therefore, the floor reaction force is controlled by the angular velocity that is the first order integral of the force or the angle that is the second order integral, and the causality concerning the angular velocity or the position of the angle and the force is There is a problem that a delay is caused due to not being satisfied. These delays are caused by, for example, a time required for converting force to angle in response time for an external force applied as a disturbance. Furthermore, since feedback control using a floor reaction force sensor is performed, there is a problem that a delay due to sensor feedback occurs. Since these delays lead to a delay in control, there is a possibility that various restrictions are added to perform a safe operation. Further, since a highly accurate floor reaction force sensor is required, there is a problem that the cost is increased. Furthermore, since inverse kinematics calculation uses an inverse matrix, there is a possibility that the solution may diverge and there is a case where a unique target value cannot be calculated.

  The present invention has been made in view of such circumstances, and does not necessarily require a floor reaction force sensor. By calculating a torque value as a control target value of an actuator used for each joint, delay is suppressed and stable. It is possible to realize attitude control, and it is possible to suppress the cost increase caused by the sensor, and because it does not require inverse kinematics calculation, the target value is uniquely determined to realize stable attitude control. It is an object of the present invention to provide a robot that can control the operation of the robot.

  A robot according to a first aspect of the present invention is a robot including a plurality of drivable coupling parts, and a plurality of actuators that drive the coupling parts based on a torque value received as a control target value, and an assumed external contact part Calculation means for calculating the torque value of each actuator that drives each connecting portion based on a target acting force set to act, and means for outputting the calculated torque value to each actuator as a control target value It is characterized by that.

  In the present invention, by controlling each actuator using the torque value as a target value, a sensor for measuring the external force is not necessarily required, and no inverse kinematic calculation is required. It is possible to calculate a unique target value by preventing the divergence of the force, and by responding to external force without breaking the causality by controlling the actuator based on the torque value indicated by the force dimension Is possible. In addition, it is possible to suppress an increase in cost caused by the sensor.

  A robot according to a second invention is characterized in that, in the first invention, the actuator is a direct-acting actuator that expands and contracts the connecting part and / or a rotary actuator that rotates the connecting part.

  The present invention can be developed in various forms using various actuators such as a direct acting actuator such as a hydraulic cylinder and a rotary motor.

A robot according to a third invention further comprises means for receiving the setting of the target acting force in the first invention or the second invention, wherein the calculating means is a torque value of each actuator based on the received target acting force. It is comprised so that it may calculate.

  In the present invention, it is possible to accept an operation from outside as the setting of the target acting force.

  According to a fourth aspect of the present invention, there is provided the robot according to the first or second aspect, wherein the means for receiving the target motion information indicating the target motion and the derivation means for deriving the target acting force to be set based on the received target motion information The calculation means is configured to calculate a torque value of each actuator based on the derived target acting force.

  In the present invention, it is possible to accept an operation from the outside as target exercise information indicating exercise such as balance, walking, and stopping.

  A robot according to a fifth invention is characterized in that, in the fourth invention, the calculating means is configured to calculate a torque value of each actuator from a target acting force based on a forward kinematic model. .

  In the present invention, based on a forward kinematic model that is easy to calculate, the target acting force is converted into a torque value, thereby reducing the calculation load required to calculate the torque value, increasing the processing speed, and suppressing delay. Is possible.

  A robot according to a sixth aspect of the present invention is the robot according to the fifth aspect, wherein the calculating means calculates the torque value of each actuator in consideration of at least one of inertia force, Coriolis force, and centrifugal force. It is characterized by being.

  In the present invention, it is possible to perform highly accurate control by using a dynamic model in addition to a forward kinematic model.

  A robot according to a seventh aspect of the present invention is the robot according to any one of the first to sixth aspects, further comprising detection means for detecting a force received from the outside, wherein the calculation means takes into account the force detected by the detection means. The present invention is characterized in that the torque value of the actuator is calculated.

  In the present invention, it is possible to perform control with higher accuracy by using a detection means such as a floor reaction force sensor for detecting a floor reaction force.

  A robot according to an eighth invention is the robot according to any one of the first to seventh inventions, wherein the contact portion is a set of a plurality of contact points, and the calculation means calculates a target acting force based on a norm minimum norm. The torque value of each actuator is calculated based on the force distributed to each contact point.

  In the present invention, the force applied to the contact points of the legs such as the toe and the heel is appropriately distributed based on the norm minimum norm, particularly the weighted norm minimum norm, and generation of a large force that cancels each other is prevented. Is possible.

  The robot according to a ninth aspect is the robot according to any one of the first to eighth aspects, wherein the calculation means calculates a torque value of each actuator based on the target acting force so as to compensate for gravity applied to the center of gravity. It is configured as described above.

  In the present invention, by calculating a torque value that compensates for gravity applied to itself, a zero-gravity state is realized under the condition of ground contact, so that the external force is relaxed by following the direction of the external force rather than repelling the external force. Is possible.

  The robot according to a tenth aspect of the present invention is the robot according to any one of the first to ninth aspects, wherein the torque value of each actuator is determined based on the restraining force that suppresses the internal motion caused by the redundant degrees of freedom related to the plurality of connecting portions. It is configured to calculate.

  In the present invention, by suppressing the internal movement with the suppression force, it is possible to prevent the operation of the joint against the intention, and to prevent adverse effects such as reaching the movable angle limit of the joint.

  A robot according to an eleventh aspect is characterized in that, in any one of the first to tenth aspects, the robot includes a plurality of legs that operate by driving the connecting portion.

  The present invention can be applied to legged robots such as bipedal walking robots.

  A robot according to a twelfth aspect of the present invention is the robot according to the first or second aspect of the present invention, wherein the biological body is configured to be wearable, and means for detecting the force received from the biological body and a target action set based on the detected force And a means for deriving a force, wherein the calculating means is configured to record so as to calculate a torque value of each actuator based on the derived target acting force.

  The present invention can be applied to an exercise assisting tool that is worn by a living organism such as a person or an animal other than a person and supports operations such as carrying heavy objects and exercising a handicapped person.

  A control device according to a thirteenth aspect of the present invention is a control device that controls the operation of a robot having a plurality of connecting portions that can be driven by an actuator, and is based on a target acting force that is set to act on an assumed external contact portion. And a means for calculating a torque value of each actuator for driving each connecting portion, and a means for outputting the calculated torque value to each actuator as a control target value.

  In the present invention, it is applied to control of the operation of the robot, and by controlling each actuator with the torque value as a target value, a sensor for measuring external force is not necessarily required, and inverse kinematics calculation is not necessarily required. It is not necessary to use an inverse matrix, it is possible to calculate a unique target value by preventing divergence of the solution, and break the causality by controlling the actuator based on the torque value indicated by the force dimension It is possible to respond quickly to external forces without any problems. In addition, it is possible to suppress an increase in cost caused by the sensor.

  The robot and the control device according to the present invention include a linear motion actuator that expands and contracts the connecting portion based on a target acting force that is set to act on an assumed external contact portion, and a rotation that rotates the connecting portion. A torque value of an actuator that drives a connecting portion such as a mold actuator is calculated, and the calculated torque value is output to each actuator as a control target value to control the actuator.

  With this configuration, in the present invention, it is not necessary to use an inverse matrix for calculation of the torque value that is the control target value, so that the solution that becomes the target value does not diverge regardless of the calculation conditions, and the unique target It is possible to calculate the value, and there are excellent effects such as being able to realize stable posture control. Moreover, by controlling the actuator based on the torque value shown in the force dimension as well as the external force, it is possible to respond quickly to the external force without breaking the causality concerning the angular velocity or the position of the angle and the force. As a result, excellent effects such as being able to perform stable posture control that is strong against disturbances are obtained. Prompt control is advantageous in performing safe control such as avoiding collisions with people and objects. And since the sensor which detects external force is not necessarily required, there exists an outstanding effect that the cost rise resulting from a sensor can be suppressed.

  The robot or the like according to the present invention can be operated from the outside by setting the target acting force received from the outside, and accepts the target motion information indicating the exercise of balance, walking, stop, etc. from the outside, By deriving the target acting force that is set based on the received target exercise information, it is possible to obtain an excellent effect such as being able to be operated from the outside.

  In particular, the robot or the like according to the present invention calculates a torque value to be distributed to each actuator from a target acting force based on, for example, a forward kinematic model using a Jacobian matrix.

  With this configuration, in the present invention, since the torque value is calculated using a forward kinematic model that is easy to calculate, the calculation load required for calculating the torque value is reduced, the processing speed is increased, and the delay is suppressed. It is possible to achieve an excellent effect. Furthermore, in addition to the forward kinematic model, it is possible to perform highly accurate control by calculating the torque value of the actuator taking into account the dynamic model such as inertia force, Coriolis force, centrifugal force, etc. There is an effect.

  The robot or the like according to the present invention can be controlled with higher accuracy by using an external force sensor that detects a force received from the outside, for example, a detection means such as a floor reaction force sensor that detects a floor reaction force. In addition, since the detection means is used as an auxiliary, it can be used even with relatively low accuracy. Therefore, it is possible to perform high-accuracy control with an inexpensive detection means. .

  The robot or the like according to the present invention generates a large force to cancel each other by optimally allocating the floor acting force to the contact points of interest such as toes and heels based on the norm minimum norm, particularly the weighted norm minimum norm. It has excellent effects such as being able to prevent. In addition, for example, when applied to a humanoid robot, various parts similar to those of humans such as balance control using the arm part are considered by considering the leg part such as the toe and the heel, and the arm part such as the elbow, the hand, and the palm as contact points. It has excellent effects such as being applicable to operation.

  A robot or the like according to the present invention realizes an operation simulating a weightless condition on the condition of grounding by providing a target acting force that guarantees gravity applied to the robot in a mathematical expression for calculating a torque value. In addition, by adopting a configuration that compensates for gravity, the external force is mitigated by performing an operation that follows the direction of the external force according to the external force, rather than repelling the external force. In addition, there is an excellent effect such that a human can apply an external force to the robot to guide the operation direction. In addition, by adopting a configuration that compensates for gravity, the program tries to maintain a state in which a force is applied to the ground from the floor contact point that is the contact portion of the robot with respect to the ground. There are excellent effects such as being able to maintain a good grounding state without changing. For example, it is easy to control the force applied to the contact portion even on an unknown uneven road surface so as to be in a stable state.

  In the case of realizing the gravity compensation of the present invention with a conventional robot that feeds back the floor reaction force detected by the floor reaction force sensor and controls the actuator with the angular velocity or angle of the joint as a target value to adjust the floor reaction force, In addition to the floor reaction force, any external force applied from the outside must be detected. It is impractical to detect an external force by arranging an infinite number of detection means in every part of the robot from both technical and economic viewpoints.

  Further, in order to accurately realize the floor reaction force, an equation of motion is derived based on an accurate model of the robot and the external environment, and an attempt to compensate all the nonlinear dynamics with the control input is partially shown in Non-Patent Document 1 described above. Has been. However, it is not practical in various aspects such as modeling error, sensor noise, and calculation cost. On the other hand, in the present invention, it is possible to compensate for only the gravity component that is less affected by modeling errors and noise. That is, the robot according to the present invention has excellent effects such that it is possible to easily realize dynamic compensation including gravitational compensation, which cannot be easily achieved by the conventional method, within a practical operating range.

  The robot or the like according to the present invention suppresses internal motion caused by redundant degrees of freedom based on a plurality of joints with a suppression force such as joint damping. When there is a redundant degree of freedom, there is an advantage that the torque value of the joint that does not contribute to the control target becomes zero or a value close to zero. There is a possibility that it may lead to an abnormality such as reaching the limit of the movable angle. In the present invention, by providing the suppression force term, it is possible to suppress the internal movement, and it is possible to prevent an unexpected joint operation and occurrence of an abnormality based on the operation.

  The robot of the present invention is not only applied to a legged robot having a plurality of legs, for example, a humanoid bipedal walking robot, but also a human being or a living organism such as an animal other than a person is attached to carry a heavy object, It can be deployed to exercise assistive tools that support movements of people with disabilities, and can also be used as a suspension system for driving wheels of vehicles that run on uneven roads. It can also be applied to various types of robots that operate in various fields such as disaster sites that cannot enter the area, volcanoes, deep seas, etc., as well as explorers that explore other celestial bodies, etc. Excellent effect. In addition to exploration, the present invention can be applied to a complex mobile robot including wheels and legs that function as a support and drive mechanism, and can be used for building work that stably performs work on rough terrain.

It is an external view which shows the robot which concerns on Embodiment 1 of this invention. It is explanatory drawing which shows typically the frame | skeleton and joint of the robot which concern on Embodiment 1 of this invention. It is a block diagram which shows the structure of the robot which concerns on Embodiment 1 of this invention. It is explanatory drawing which shows typically the coordinate system which concerns on the robot which concerns on Embodiment 1 of this invention, and control of a robot. It is explanatory drawing which shows typically the floor acting force which concerns on the robot which concerns on Embodiment 1 of this invention, and a robot. It is explanatory drawing which shows typically the virtual contact force which concerns on the robot which concerns on Embodiment 1 of this invention, and a robot. It is a flowchart which shows the process of the drive control apparatus with which the robot of this invention is provided. It is an external view which shows typically the bending movement of the robot which concerns on Embodiment 1 of this invention. It is a graph which shows the time-dependent change of the value which shows the operation | movement of bending and stretching movement in the simulation experiment of the robot which concerns on Embodiment 1 of this invention. 6 is a graph showing temporal changes in the center of gravity and ZMP during bending and stretching movements in the simulation experiment of the robot according to Embodiment 1 of the present invention. It is an external view which shows typically the leg raising motion of the robot which concerns on Embodiment 1 of this invention. 7 is a graph showing a change with time of the position of the leg during the leg raising exercise in the simulation experiment of the robot according to the first embodiment of the present invention. It is a graph which shows a time-dependent change of the value which shows leg raising exercise | movement in the simulation experiment of the robot which concerns on Embodiment 1 of this invention. It is a graph which shows the time-dependent change of the value which shows walking motion in the simulation experiment of the robot which concerns on Embodiment 1 of this invention. It is an external view which shows the robot which concerns on Embodiment 2 of this invention. It is an external view which shows the robot which concerns on Embodiment 3 of this invention. It is an external view which shows the robot which concerns on Embodiment 4 of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Robot 10 Leg part 2 Connection part 20 Actuator 3 Drive control apparatus 30 Control means 31 Recording means 32 Storage means 33 Measuring means 34 Output means 35 Input means 4 Drive mechanism 5 Detection mechanism

Embodiment 1 FIG.
Hereinafter, the present invention will be described in detail with reference to the drawings illustrating embodiments thereof. FIG. 1 is an external view showing a robot according to Embodiment 1 of the present invention, and FIG. 2 is a block diagram schematically showing a skeleton and joints of the robot according to Embodiment 1 of the present invention. The first embodiment exemplifies a form in which the present invention is applied to a legged robot that includes at least a pair of leg portions and performs various operations such as walking, bending and stretching, and leg raising.

  1 and 2, reference numeral 1 denotes a robot. The left and right legs 10, 10 of the robot 1 are provided with connecting portions 2, 2... Such as joints on the waist, knees, and ankles. ,... Are driven by actuators 20, 20,. By being driven by the actuators 20, 20,..., The connecting portions 2, 2,... Can be bent in a plurality of directions such as front and rear, left and right. The robot 1 includes not only the legs 10 and 10 but also actuators 20, 20,... That drive the joints 2, 2, etc. at various locations such as the neck, chest, shoulders, elbows, and wrists. ing.

  As the actuator 20, various actuators such as a servo motor and a hydraulic motor can be used as long as the actuator 20 has a function of receiving a torque value as a control target value as a drive signal and performing control based on the received torque value. It is. For example, in a servo motor that has a drive circuit capable of current control and generates torque proportional to the current, the torque value input as the control target value is multiplied by a torque constant determined by the gear ratio, and a command is sent to the drive circuit. Thus, torque control for generating the input torque is realized. In particular, by providing a torque sensor in the connecting portion 2 and feeding back the value detected by the torque sensor to the drive circuit, highly accurate torque control is possible. Further, not limited to the rotary type, it is also possible to use a direct acting actuator 20 such as a hydraulic cylinder. That is, the number and arrangement of the connecting portions 2, 2,... Of the robot 1 shown in FIG. 2 is merely an example, and the connecting portions including various types of actuators 20, 20,. 2, 2, ... can be arranged in various places.

  FIG. 3 is a block diagram showing a configuration of the robot 1 according to the first embodiment of the present invention. The robot 1 controls the actual torque based on the drive control device 3 that outputs signals to the actuators 20, 20,... That drive the connecting portions 2, 2,. .. Such as servo amplifiers for driving the actuators 20, 20,... To drive the actuators 20,..., And various detection mechanisms 5, 5,. It has. The drive mechanisms 4, 4,... May be mounted as drive means provided in the drive control device 3, or an intelligent motor integrated with the actuator 20 may be mounted.

  The angle sensor is a sensor that detects the angle of each joint provided as the connecting portions 2, 2,..., And for example, a sensor such as an analog potentiometer or a digital rotary encoder is used. The attitude sensor is a sensor that detects the absolute attitude of the robot 1 in the inertial coordinate system. For example, a gyro sensor attached to the body of the robot 1 is used. As the external force sensor, for example, a floor reaction force sensor attached to the sole of the robot 1 is used. Note that the robot 1 of the present invention does not necessarily require an external force sensor, but by using the external force sensor together, it is possible to control the force applied to the outside according to the actual contact state. That is, it is effective in determining whether or not to distribute the acting force to be applied to each contact point according to the actual contact situation. Furthermore, detection of an external force by an external force sensor is also effective when setting a weight according to the state of friction with the outside.

  The drive control device 3 includes a control unit 30 such as a CPU that performs various calculations such as calculation of a control target value, a recording unit 31 such as a ROM, EPROM, and hard disk that records information such as programs and data required for control, A storage means 32 such as a RAM for temporarily storing data generated by the execution of the program, a measurement means 33 for receiving input of various signals indicating detection results from the detection mechanisms 5, 5,. An output unit 34 that converts a signal into a format corresponding to the drive mechanism 4 and outputs the signal and an input unit 35 that receives an input from the outside are provided.

  The signal output from the drive control device 3 to each actuator 20, 20,... Via the drive mechanism 4 is a control signal indicating a torque value as a control target value, for example. The actuators 20, 20,... That drive the connecting portions 2, 2,... Operate based on a control signal such as a torque value input from the drive control device 3. The input means 35 is a mechanism for receiving target motion information indicating a target motion such as balance, walking, or stopping of the robot 1 or a target acting force for performing the target motion. A command including information such as target exercise information transmitted from the controller is received. Further, an operation unit such as a keyboard, various buttons, or a switch that directly receives a command from the operator may be used as the input unit 35. When the drive control device 3 accepts the target motion information, it derives a target acting force to be set and a target zero moment point (hereinafter referred to as ZMP: Zero Moment Point) from the target motion, and derives the target Based on the acting force and the target ZMP, the robot 1 is controlled as described later. In addition, when the drive control device 3 receives the target acting force, the drive control device 3 controls the robot 1 based on the received target acting force and the target ZMP derived from the target acting force.

  In addition, the object which contact parts, such as a grounding point demonstrated in this application, contact is not restricted to the ground, but refers to the general external environment where the robot 1 applies its own weight to support itself. Specifically, in addition to the ground and the floor, it also includes external environments other than planes such as sloped floors and steps with steps, etc. In addition to the soles of legs, knees, hands, arms, etc. When taking a posture to support itself using a part other than the sole of the foot, it includes external environments such as handrails and walls. Therefore, when taking a posture such as a crawl that supports itself using the upper body, a hand portion (front leg portion) such as a palm can also be a contact portion. Similarly, the floor reaction force described in the present application is not necessarily limited to the reaction force received from the floor, but indicates the reaction force received from the external environment in contact with the contact portion.

Next, control of the robot 1 of the present invention will be described. FIG. 4 is an explanatory diagram schematically showing the robot 1 according to the first embodiment of the present invention and a coordinate system related to the control of the robot 1. CoM in FIG. 4 is the center of mass of the robot, and the position vector r defined by the following equation 1 in the three-dimensional coordinate system Σ using the x, y, and z coordinates. Define in _C. In the coordinate system Σ shown in FIG. 4, the center of gravity CoM is indicated by a position vector r _C with a point outside the robot 1 as a reference point. However, a coordinate system that takes the reference point in the robot 1 is defined. May be.

r _C = [x _C , y _C , z _C ] ^T ∈ R ³ (Equation 1)
Where r _C : position vector of the center of gravity CoM
^T : Transpose operator for vector or matrix
R ³ : three-dimensional number vector space

In order to simplify the explanation, first, the situation where the robot 1 is in contact with an external environment at an arbitrary point on the sole will be described. A position vector from the center of gravity CoM of the contact point is indicated as r _P , and is defined by the following Expression 2 using a three-dimensional coordinate system based on x, y, and z coordinates.

r _P = [x _P , y _P , z _P ] ^T ∈ R ³ (Equation 2)
Where r _P : contact point position vector

  The force applied to the center of gravity CoM can be expressed by the following Equation 3 using other external forces such as gravity, a floor reaction force received from the contact point, and a disturbance. In addition, when the robot 1 is in contact with the outside, here the floor, only at this contact point, the contact point coincides with ZMP.

Md ² r _C / dt ² = Mg + f _R + f _E (Formula 3)
Where M is the mass of the robot 1
d ² r _C / dt ² : acceleration applied to the center of gravity
g: Gravity acceleration indicated as [0, 0, −9.81] ^T
f _R : floor reaction force with f _R ∈R ³
f _E : external force other than gravity and floor reaction force where f _E ∈R ³

  The position vector of the points such as CoM and ZMP shown in Expression 2 can be derived based on the detection values detected by the angle sensor and the attitude sensor that are the detection mechanism 5. The drive control device 3 has a floor applied force (GAF: Ground Applied Force) that is the same in magnitude and opposite to the floor reaction force from the robot 1 to the contact point. ) Is set to a desired value according to the task target of the operation of the robot 1, and the robot 1 is controlled so that the determined floor action force is applied from the robot 1 to the contact point. By this control, the robot 1 can perform various follow-up operations such as balancing itself and performing stable posture control. The floor acting force is defined by the following equation 4 using a three-dimensional coordinate system based on x, y, and z coordinates.

f _P = [f _xP , f _yP , f _zp ] ^T = −f _R ∈R ³ (Formula 4)
Where fP: floor force

  The drive control device 3 drives each joint 2, 2,... Based on the forward kinematic model specified by the Jacobian matrix, as shown in the following equation (5). The torque values of the actuators 20, 20,... Are converted, and the converted torque values are output to the actuators 20, 20,. And each actuator 20, 20, ... operates based on the torque value which received the input.

  As shown in Equation 5, the torque value that is the control target value of each actuator 20, 20,... Of each joint can be calculated using a transpose matrix. Further, since the inverse matrix is not used in the calculation process, the solution does not diverge. When the actual floor reaction force matches the force in the reverse direction of the target floor action force calculated by Equation 5, the robot 1 of the present invention can achieve a desired motion with a very simple calculation. It is also possible to perform various following operations such as balancing itself and performing stable posture control. The inventors of the present application have already examined and disclosed whether or not the floor reaction force actually matches the force in the direction opposite to the floor action force.

Since the robot 1 performs various operations such as walking, it is not always in contact with the floor at one point of the ZMP. Then, control of the robot 1 of the present invention under the situation where it is in contact with the floor at a plurality of contact points will be described next. FIG. 5 is an explanatory diagram schematically showing the robot 1 according to the first embodiment of the present invention and the floor acting force according to the robot 1. In the robot 1 illustrated in FIG. 5, the floor acting force f _P is shown as a vector from the center of gravity CoM to the direction of ZMP. In FIG. 5, a set of points in contact with an external environment such as a floor is considered. The contact part of the leg 10 is regarded as a set of a plurality of contact points that contact the floor surface, attention is paid to specific a contact points in the contact part of the leg 10, and the position vector of each contact point is It is defined by the following expression 6 using a three-dimensional coordinate system based on x, y and z coordinates.

r _S = [r _S1 , r _S2 ,..., r _Sa ] ^T ∈ R ^3a (Formula 6)
Where r _Sj : position vector of each contact point Sj (j = 1, 2,..., A)

Then, paying attention to a specific contact point in the set of contact points, and considering the floor acting force as a resultant force of the force applied to the focused contact point, the floor acting force f _P is defined by the following Expression 7. Can do.

For example, in the robot 1 shown in FIG. 5, a portion corresponding to the sole of the toe portion having a rectangular parallelepiped shape from the heel to the toe is used as a contact portion of the leg portion. Right tip S ₁ , left tip S ₂ , right rear end S ₃ and left rear end S ₄ of the foot tip, and right tip S ₅ , left tip S ₆ , right rear end S ₇ of the foot tip of the left leg The left rear end S ₈ is a contact point of interest.

  Note that the number of contact points of the robot 1 that performs movements such as walking is not always constant. For example, when the robot 1 performs a walking motion, since the free leg is not in contact with the floor surface, the floor acting force is distributed only to the contact point on the support leg side. Further, for the free leg that is not in contact with the floor surface, the virtual contact point is defined as shown in Equation 8 as in the case of the support leg.

r _F = [r _{F 1} , r _{F 2} ,..., r _Fb ] ^T ∈ R ^3b (Equation 8)
Where r _Fj : position vector of each contact point F _j (j = 1, 2,..., B)

  These virtual contact points use the coordinates of a point that actually contacts the external environment or a point that does not contact the external environment but moves in space. For example, as shown in FIG. 5, in the robot 1 that is in contact with the floor with both legs 10 and 10, a = 8 and b = 0, but when one leg 10 becomes a free leg, a = 4 , B = 4. Further, when only a part of the leg portion 10 such as a toe or a heel is in contact with the external environment, it further changes. Assuming a virtual contact point on the palm as well as the sole, b = 8. Further, the x-coordinate and y-coordinate of the ZMP position vector can be expressed by the following equations 9 and 10. When the robot 1 is in contact with the external environment on a flat floor surface, the z coordinate of the ZMP position vector is zero.

  In addition, the above-mentioned Formula 7, Formula 9, and Formula 10 can be put together into the following Formula 11 as a force relational expression in the Z direction. Further, the following formulas 12 and 13 are assumed as the force relationship in the X direction and the Y direction. This is a treatment for assigning a correspondingly large horizontal contact force, that is, a frictional force, to a contact point having a larger force in the Z direction.

  Further, when the floor acting force with respect to the floor reaction force is distributed to each contact point, the target force to be distributed to each contact point is optimally distributed based on the norm minimum criterion using the following formulas 14, 15, and 16. To be decided. This prevents the generation of large forces that cancel each other. When the robot 1 takes a posture such as a heel or toe standing, standing upright, or all fours, a floor reaction force is distributed to each contact point according to a contact state based on the posture. That is, by using the following Expression 14, Expression 15, and Expression 16, the target action force that is applied to each contact point is calculated based on the target ZMP and the target floor action force.

It is also possible to set arbitrary weights such as 80% for the toe of the right leg and 20% for the heel of the left foot with respect to the target acting force applied to each contact point. Specifically, the matrix A _i ^# used in Expression 14, Expression 15, and Expression 16 is defined using an inverse matrix of an arbitrary weight coefficient matrix as shown in Expression 17 below, and Expression 14 By modifying Equations 15 and 16, it is possible to distribute the target acting force based on the weighted norm minimum criterion.

A _i ^# = (A _i ^T W ⁻¹ A _i ) ⁻¹ A _i ^T W ⁻¹ (Expression 17)
W: Weight coefficient matrix

  When the weight coefficient matrix is used, the target contact is set so that the value of the optimum evaluation function shown as the product of the transposed matrix of the target contact force applied to the contact point S, the weight coefficient matrix, and the matrix indicating the target contact force is minimized. Force is defined. Since the target contact force is a row matrix and the transpose matrix is a column matrix, the value of the optimum evaluation function is a scalar quantity.

  Then, the drive control device 3 is applied to each contact point so that the floor acting force is optimally distributed based on the norm minimum norm based on the above-described Expression 7, 9, 9 and 10, 14, 14, 15 and 16. Based on the forward kinematic model defined by the Jacobian matrix, the actuator 20 that drives each joint as shown in the following equation 18 is based on the floor action force that is derived from the force and distributed to the force applied to each contact point. , 20,..., And the converted torque values are output as control target values to the actuators 2, 2,. And each actuator 20, 20, ... operates based on the torque value which received the input.

  Each actuator 20, 20,... Operates based on a torque value that has received an input, whereby the robot 1 of the present invention performs a stable operation during exercise such as walking.

  Next, a control method for suppressing internal movement in the robot 1 of the present invention will be described. When the torque values of the actuators 20, 20,... Are calculated using Equation 5 or Equation 18 for the robot 1 having a large number of joints 2, 2,. Therefore, the target torque of some joints takes zero or an approximate value of zero, and there is a possibility that an internal motion in which the joints operate unexpectedly occurs. Therefore, the drive control device 3 of the present invention calculates the torque value of each actuator 20, 20,... Using the following equation 19 provided with a suppression force term that suppresses internal motion.

  Expression 19 is an expression in which the suppression force term is provided in the above-described Expression 18, the first term on the right side is Expression 18, and the second term is the suppression force term. The simplest suppression force term is one in which damping is provided for each joint, and is expressed by the following equation 20 in which the damping coefficient for each joint is shown as a matrix.

As shown in Expression 20, by setting an arbitrary friction coefficient for each actuator 20, 20,... As a damping coefficient, the internal movement based on the free movement of the actuators 20, 20,. . τ _a is a value that can be set to an arbitrary value as necessary to correct the target value of the torque value, and is used, for example, when it is desired to specify a local posture and angle. In this way, the drive control device 3 controls based on the torque values of the actuators 20, 20,. , Perform the action according to the operator's intention. In addition, you may make it use the formula which provided the suppression force term in the above-mentioned Formula 5. FIG.

  Next, a control method that operates flexibly with respect to external force in the robot 1 of the present invention will be described. The drive control device 3 according to the present invention controls the robot 1 by providing a term that adds gravity to the target value of the floor acting force in order to compensate for the gravity applied to the center of gravity of the robot 1 as shown in the following Expression 21. . By performing control to compensate for gravity, the robot 1 realizes an operation simulating a weightless condition on condition that the robot 1 is in contact with the floor. That is, the robot 1 of the present invention does not repel the external force but performs an operation of following the direction of the external force according to the external force.

  In Equation 21, the arbitrarily set force is a target acting force that can be arbitrarily set to control the operation of the robot 1. For example, information that causes the robot 1 to perform an operation of lowering the center of gravity is used as target motion information. If accepted, an arbitrarily set force is set as the downward force.

  Based on Equation 21 and Equation 3 above, the following Equation 22 can be derived.

Md ² r _C / dt ² = f _D + f _E + f _u (Equation 22)
Where f _D : restraining force term based on the second term on the right side of Equation 19

  Considering Equation 21 as shown in Equation 22, the apparent influence of gravity can be eliminated. The drive control device 3 controls the robot 1 based on the target torque value of the actuators 20, 20,. 1 is not repelled by an external force, but operates according to the external force. Therefore, the external force is relaxed and safety at the time of a collision with a person and an object is improved. For example, when a person applies an external force to the robot 1, It is possible to guide.

  The arbitrary setting force shown in Expression 22 is a target acting force derived from information received by the robot 1 as target motion information as described above. Various methods for deriving the target action force have been proposed. For example, the target action force is calculated from the target position and target speed of the center of gravity of the robot 1 in the three-dimensional space indicated by the target motion information using the following Expression 23. Force can be derived.

  Equation 23 is a derivation method using a simple linear feedback law, but the drive control device 3 of the present invention derives a target acting force that is treated as an arbitrarily set force by various derivation methods other than Equation 23. Is possible. Furthermore, the drive control device 3 of the present invention can derive the target ZMP based on the following formula 24 from the target acting force that is an arbitrarily set force.

  When the target ZMP derived by Equation 24 is in contact with the ground, that is, not in contact with the external environment, the closest contact point among the infinite number of contact points in contact with the external environment is re-derived as the target ZMP, A target acting force is derived based on the derived target ZMP. However, when the horizontal force of the target acting force is zero, that is, when only the gravity compensation or the vertical movement of the center of gravity is performed, both the x coordinate and the y coordinate viewed from the center of gravity CoM of the target ZMP are zero.

  Next, in the robot 1 of the present invention, as a method for controlling the floor reaction force with higher accuracy, a method of performing inverse dynamics calculation in consideration of nonlinear terms such as an inertial force term, a Coriolis force term, a centrifugal force term, etc. with respect to the robot 1 There is. A non-linear term for the robot 1 is shown in the barycentric coordinate system as shown in Equation 25 below.

  From the above equation 25, the floor acting force can be calculated using the following equation 26.

  The drive control device 3 can calculate the torque value of each actuator 20, 20,... Of each joint from the target value of the floor action force by solving Equation 26 for u ^ (^ symbol above u). it can. However, since it is difficult to accurately calculate all the nonlinear terms, particularly terms including the square of the noisy speed, an estimated value calculated using an appropriate filter may be used. Since gravity compensation is performed as described above, high followability can be expected even when an error is included in dynamics calculation other than gravity compensation.

Next, in the robot 1 of the present invention, a whole body motion control method when various operations are performed using the extremities such as the hand portion and the leg portion 10 that are not in contact with the external environment will be described. FIG. 6 is an explanatory diagram schematically showing the robot 1 according to Embodiment 1 of the present invention and the virtual contact force related to the robot 1. The drive control device 3 performs, for example, PD control as a target motion such as the motion indicated by the target motion information that has received the input, for example, when performing a reaching motion that moves the arm or leg 10 to a desired position in space. As a result, a virtual contact point of the robot 1 is assumed, and a target virtual contact force to be applied to each virtual contact point is derived. FIG. 6 shows a virtual contact point r _Fj and a target virtual contact force f _Fj based on the target motion. The drive control device 3 can calculate the torque values of the actuators 20, 20,... Based on the calculated target virtual contact force f _Fj based on, for example, the following equation 27.

However, since the sum of the virtual contact forces determined in this way becomes a reaction force with respect to the center of gravity, there is a possibility that the balance will be adversely affected. So this force must be compensated. For this purpose, a value obtained by subtracting the total sum of the virtual acting forces from the target floor acting force of Expression 21 is again used as the target acting force. Note that the concept of the virtual contact point r _Fj shown in FIG. 6 is introduced and calculation is not performed using the Jacobian matrix shown in Expression 27, but the contact point r _Sj described above is assumed to be in contact. The calculation may be performed using a mathematical expression related to the point r _Sj , and in that case, the calculation using the Jacobian matrix shown in Expression 27 is not necessary.

  When the robot 1 performs a motion such as a walking motion, the contact force of the support leg can be used not only to generate a floor acting force corresponding to a translational force but also to generate a floor acting moment corresponding to a rotational force. Because there are a plurality of contact forces, a moment can be generated by combining them in pairs. For example, the orientation of the robot 1 can be controlled by appropriately generating a moment with respect to the contact portion.

  Next, control processing of the robot 1 using the drive control device 3 according to Embodiment 1 of the present invention will be described. FIG. 7 is a flowchart showing processing of the drive control device 3 provided in the robot 1 according to Embodiment 1 of the present invention. An operator who operates the robot 1 drives the robot 1 with target motion information indicating a motion such as a balance motion that maintains a posture so as not to fall down, a bending / extending motion, and a reaching task motion that causes a hand or a tip to reach a target. Input to the control device 3. The input of the target motion information is performed, for example, when the operator operates the controller for the robot 1 and transmits the target motion information to the drive control device 3 from the controller through wired or wireless communication. The drive control device 3 provided in the robot 1 of the present invention receives input of target motion information by the input means 35 under the control of the control means 30 (step S1). In step S <b> 1, for example, the input unit 35 receives target exercise information transmitted from the controller through wireless communication, thereby receiving input. The drive control device 3 converts information such as the flexion / extension motion indicated by the received target motion information into information indicating the motion of each part of the robot 1 based on the setting indicating the correspondence relationship of the motion recorded in advance. Then, the motion related motion is recognized. Note that instead of the target motion information, the target acting force and the target ZMP may be input to the robot 1, and in step S1, the input unit 35 may receive the input target acting force and the target ZMP.

  The drive control device 3 included in the robot 1 derives a target acting force and a target ZMP according to the target motion indicated by the received target motion information under the control of the control unit 30 (step S2). In step S2, for example, a target acting force, in this case, a floor acting force is derived based on Equation 23 described above, and a target ZMP is derived based on Equation 24. In addition, as shown in the description using Formula 8 or FIG. 6, a virtual contact point may be assumed and the target ZMP and the target acting force may be derived. As described above, the method for deriving the target acting force may use a method other than Equation 23, and the method for deriving the target ZMP may use a method other than Equation 24. In calculating the target acting force, if necessary, gravity compensation based on the above-described equation 21, arbitrary setting force based on the equation 22, and nonlinear forces such as inertial force, Coriolis force, and centrifugal force based on the equations 25 and 26 are used. The power related to the term is taken into account. The derived target acting force and target ZMP are stored in the storage means 32 as set values. When the input means 35 accepts the target applied force and the target ZMP in step S1, the process of step S2 is a process of deriving data in a format that can be handled in the subsequent processes from the received target applied force and target ZMP. It becomes. Alternatively, only the target acting force may be received and the target ZMP may be derived from the target acting force using Equation 24.

  And the drive control apparatus 3 with which the robot 1 of this invention is equipped is based on the above-mentioned Formula 5, Furthermore, Formula 7, Formula 9, and Formula 10, Formula 14, Formula 15 and Formula 16 by control of the control means 30. A force applied to each contact portion is derived so that the set acting force is optimally distributed based on the norm minimum norm (step S3), and the applied force is distributed to the force applied to each contact portion based on Equation 18 described above. The torque values are calculated as control target values for the actuators 20, 20,... That drive the connecting portions 2, 2,.

  Then, the drive control device 3 provided in the robot 1 of the present invention controls the actuators 20 through the drive mechanisms 4, 4,. 20,... (Step S5). Each actuator 20, 20,... Receives a torque value as a control target value and operates based on the received torque value.

  Next, a simulation experiment result regarding the robot of the present invention will be described. FIG. 8 is an external view schematically showing the bending and stretching motion of the robot according to the first embodiment of the present invention. FIG. 8 shows the bending / extending motion assumed as a simulation experiment of the present invention. The robot is shown in FIG. 8 (a), FIG. 8 (b), FIG. 8 (c), FIG. 8 (d), and FIG. The operation is repeated in the order of e).

FIG. 9 is a graph showing a change with time of a value indicating the bending / extending motion in the simulation experiment of the robot according to the first embodiment of the present invention. FIG. 9A is a graph showing a change with time in the height Z _C of the center of gravity of the robot in the bending and stretching movement shown in FIG. 8, and FIG. 9B is a graph showing the trunk of the robot based on the upright position. it is a graph showing temporal changes in the longitudinal direction of the angle phi _p and the left-right direction of the angle phi _r. As shown in FIG. 9 (a), the robot is controlled to perform a bending / extending motion in which the height of the center of gravity changes based on a sine curve, and as shown in FIG. 9 (b), the robot The body is controlled to perform bending and stretching movements in which the front and rear and left and right angles of the trunk change based on a sine curve. In the simulation experiment shown in FIG. 9, the target value of each joint related to control is arbitrarily set within the movable range of each joint.

FIG. 10 is a graph showing temporal changes in the center of gravity and ZMP during bending and stretching movements in the simulation experiment of the robot according to Embodiment 1 of the present invention. FIG. 10A is a graph showing the time-dependent change of the x-coordinate x _C of the center of gravity of the robot and the time-dependent change of the x-coordinate x _P of the ZMP in the bending and stretching movement shown in FIG. 10 (b) is a graph showing changes with time and aging of the y-coordinate y _P of ZMP y coordinate y _C of the center of gravity whose origin is the center of the legs. The x coordinate is a coordinate indicating the front-rear direction of the robot, and the y coordinate is a coordinate indicating the left-right direction of the robot. FIG. 10 shows changes in the center of gravity and ZMP when the bending and stretching movement shown in FIG. 9 is performed. 9 and 10 that the ZMP swings back and forth and right and left while the robot is bending and extending, but it can be seen that the center of gravity is substantially constant. This is because the robot of the present invention is controlled so that the center of gravity is stabilized by applying a load to the front, rear, left and right during bending and stretching movements.

  FIG. 11 is an external view schematically showing the leg raising motion of the robot according to the first embodiment of the present invention. FIG. 11 shows a leg raising motion assumed as a simulation experiment of the present invention. The robot raises one leg to the front right of itself as shown in FIG. The operation of lowering the legs raised in the order of FIG. 11 (c) and FIG. 11 (d) is repeated. In the simulation experiment shown in FIG. 11, the target value of each joint related to the control is arbitrarily set within the movable range of each joint.

FIG. 12 is a graph showing temporal changes in the position of the legs during the leg raising exercise in the robot simulation experiment according to the first embodiment of the present invention. FIG. 12 is a graph showing temporal changes in the x-coordinate x _F , y-coordinate y _F, and z-coordinate z _F of the toe on the free leg side with the sole of the supporting leg as the origin in the leg raising motion shown in FIG. . As shown in FIG. 12, the robot performs a leg raising motion so that the position of the foottip moves based on a smooth curve.

FIG. 13 is a graph showing a change with time of a value indicating the leg raising motion in the simulation experiment of the robot according to the first embodiment of the present invention. FIG. 13A is a graph showing temporal changes in the x-coordinate x _C and y-coordinate y _C of the center of gravity of the robot with the sole of the support leg as the origin in the leg raising motion shown in FIG. b) is a graph showing a change with time of the z-coordinate z _C of the center of gravity, and FIG. 13C is a graph showing the trunk angle Φ _p , the horizontal angle φ _r, and the yaw axis posture angle φ _y . It is a graph which shows a time-dependent change. FIG. 13 shows changes in various values when the leg raising exercise shown in FIG. 12 is performed, and the robot of the present invention can perform stable control even during the leg raising exercise.

FIG. 14 is a graph showing a change with time of a value indicating walking motion in the simulation experiment of the robot according to Embodiment 1 of the present invention. FIG. 14A shows changes over time of the velocity dx _C / dt in the x coordinate direction, the velocity dy _C / dt in the y coordinate direction, and the velocity dz _C / dt in the z coordinate direction of the center of gravity of the robot. (B) is a graph showing temporal changes in the front-rear direction angle φ _p , the left-right direction angle φ _r, and the yaw axis posture angle φ _y of the trunk of the robot. The graph shown in FIG. 14 shows the operation when the robot of the present invention performing gravity compensation control receives an external force during walking as the speed of the center of gravity and the inclination of the trunk. The robot receives an external force of 1500 N backward for 0.1 second 2 seconds after the start of the experiment, receives an external force of 1500 N forward for 0.1 second after 4 seconds, and receives an external force of 500 N rightward for 0.1 second after 5 seconds. 6 seconds later and receiving an external force of 500 N for 0.1 seconds. As shown in FIG. 14, the robot does not repel external force but continues walking motion without falling down according to the external force.

  The robot of the present invention is not limited to the above-described example, and can be developed into various forms for controlling the actuators of each joint included in the robot based on the torque value calculated from the floor acting force. For example, in the first embodiment, a robot having a pair, that is, two legs has been described. However, the present invention is not limited to this, and there are a large number of robots such as four, six, and the like that perform a handstand walking with arms. It can be developed in various forms, for example, applied to a robot having a leg portion.

Embodiment 2. FIG.
The second embodiment is an embodiment in which the robot of the present invention is applied to a motion assisting device. FIG. 15 is an external view showing a robot according to Embodiment 2 of the present invention. In FIG. 15, reference numeral 1 denotes a robot according to Embodiment 2 of the present invention, and the robot 1 is configured to be worn by a person. In FIG. 15, a person wearing the robot 1 that is an exercise assisting device is shown. This shows the situation where heavy objects are being transported. The robot 1 has human arms, trunks, and legs by means of rod-like auxiliary exoskeletons that are provided with extendable connecting parts 2, 2,... Driven by direct acting actuators 20, 20,. Is helping to exercise. A drive control device 3 including a power source of the robot 1 is attached to the back of the person.

  The configuration of the robot 1 is substantially the same as the configuration of the first embodiment shown in the block diagram of FIG. 3, and includes the connecting portions 2, 2,..., The actuators 20, 20,. A drive mechanism and a detection mechanism are provided. The drive control device 3 includes control means, recording means, storage means, measurement means, output means, and input means.

  The processing by the drive control device 3 of the robot 1 according to the second embodiment is substantially the same as that of the first embodiment described with reference to FIG. 7, and the input means is a process corresponding to step S1 in FIG. The force in the person's arm, torso and leg is detected, and the detected force is received as input of the target exercise information.

Then, as a process corresponding to step S2, the drive control device 3 of the robot 1 derives the target ZMP and the target acting force from the target motion information based on the detected force by the various calculation methods described in the first embodiment. As a process corresponding to step S3, a force applied to each contact portion is derived so that the derived acting force is optimally distributed based on the norm minimum norm. In the example shown in FIG. 15, the robot 1 of the present invention derives the floor action force distributed in the direction of each contact point indicated by the position vectors r _S1 , r _S2 , r _S3 , and r _S4 from the center of gravity CoM, and weights. The acting force distributed in the direction of each virtual contact point indicated by the position vectors r _F1 , r _F2 , r _F3 in order to lift the object is derived.

  Further, the drive control device 3 of the robot 1 calculates a torque value as a control target value of each actuator 20, 20,... As a process corresponding to step S4, and each calculated control target as a process corresponding to step S5. The value is output to each actuator 20, 20,. By such processing, a person wearing the robot 1 configured as an exercise assisting device can carry a heavy object with a small force as shown in FIG. Since the input means has a function of detecting the force of the person, it is possible to share part of the functions of the detection mechanism and the input means.

  The control of the drive control device 3 of the robot 1 will be further described in detail. In step S2, the target ZMP to be calculated is set to be the center of gravity projection point obtained by projecting the center of gravity CoM on the ground, and by performing the gravity compensation described in the first embodiment, the person wearing the robot 1 If it is a simple motion such as movement, the robot 1 can be easily operated without being aware of the operation method.

The drive control device 3 of the robot 1 can be controlled in various modes according to the purpose. First, the balance mode for automatically maintaining the balance in the horizontal direction will be described. In the balance mode, the target ZMP is set so that the center-of-gravity projection point of the robot 1 comes to the center of gravity of the contact surface that is a contact portion that contacts the ground. By performing control based on such settings, a person wearing the robot 1 as an exercise assisting device can perform a transport operation without being particularly conscious of balance even when transporting heavy objects. Further, as the setting of the drive control device 3, when the weighting coefficient matrix, Equation 23 is used, parameters relating to acceleration / deceleration such as constants K _PC and constants K _DC indicating feedback gain, PID values for the actuators 20, 20,. By doing so, it is possible to control the followability to the target ZMP and to control the balance strength.

  Next, a movement assist mode for assisting acceleration / deceleration of the center of gravity of the person in the three-dimensional space will be described. In the movement assist mode, the actual floor reaction force and ZMP are detected or calculated based on the detection result by the detection mechanism, and the center of gravity projection point calculated based on the floor reaction force and the ZMP distance are used as acceleration / deceleration parameters to be assisted. Use. By performing control using such parameters, a person wearing the robot 1 as an exercise assisting device can easily perform an operation of moving a heavy object from right to left in an upright state, for example. Further, as the setting of the drive control device 3, the strength of movement assistance can be controlled by multiplying the acceleration / deceleration parameter by the control target related to the followability.

  Next, an autonomous mode that does not use human power for control will be described. In the autonomous mode, it operates based on set values such as target motion information and / or target acting force received by the input means. In the autonomous mode, the person wearing the robot 1 can completely remove the power and only sets the target motion information such as the target moving speed and the moving point. The balance mode, the movement assist mode, and the autonomous mode described above can be made to function independently, and the operator assigns the weight to the torque value calculated in each mode, with the parameter relating to the strength of the mode as a weight. It is also possible to function simultaneously by superimposing the multiplied ones.

  In Embodiment 2 described above, a robot is used as an exercise assisting tool that can be worn by a person. However, the present invention is not limited to this and may be worn by a living organism other than a person. It is also possible to suppress the force related to the exercise and assist the work that requires fine control of the force.

Embodiment 3 FIG.
In the third embodiment, the robot of the present invention is applied to a vehicle. FIG. 16 is an external view showing a robot according to Embodiment 3 of the present invention. In FIG. 16, reference numeral 1 denotes a robot according to Embodiment 3 of the present invention, and the robot 1 is configured as a three-wheel buggy type vehicle having two front wheels and one rear wheel. The front wheels and the rear wheels are all drive wheels, and a rod-like body provided with extendable connecting portions 2, 2,... Driven by direct acting actuators 20, 20,. Is used. Other components such as a drive control device are incorporated in the vehicle body.

  The configuration of the robot 1 is substantially the same as the configuration of the first embodiment shown in the block diagram of FIG. 3, and the connecting portions 2, 2,... And the actuators 20, 20,. And a detection mechanism. The drive control apparatus includes a control unit, a recording unit, a storage unit, a measurement unit, an output unit, and an input unit.

  The processing by the drive control device for the robot 1 according to the third embodiment is substantially the same as that of the first embodiment described with reference to FIG. 7. As processing corresponding to step S1 in FIG. A passenger's operation is received as input of target motion information from a steering mechanism such as a vehicle handle, an acceleration mechanism such as an accelerator, and a braking mechanism such as a brake.

Then, as a process corresponding to step S2, the drive control device of the robot 1 derives the target ZMP and the target acting force from the received target motion information by the various calculation methods described in the first embodiment, and proceeds to step S3. As a corresponding process, a force applied to each contact portion is derived so that the derived acting force is optimally distributed based on the norm minimum criterion. In the example shown in FIG. 16, the robot 1 of the present invention derives the floor acting force distributed in the direction of each contact point indicated by the position vectors r _S1 , r _S2 and r _S3 from the center of gravity CoM.

  Further, as a process corresponding to step S4, the drive control device of the robot 1 calculates a torque value as a control target value of each actuator 20, 20,..., And as a process corresponding to step S5, each calculated control target value. Are output to the respective actuators 20, 20,. By such processing, the robot 1 configured as a vehicle such as a three-wheel buggy determines the posture corresponding to the vertical movement and steering during traveling, disperses the weight, and elastically absorbs the impact and vibration from the road surface. Suspend the vehicle body and passengers stably.

  In the third embodiment, a robot is used as a three-wheel buggy. However, the present invention is not limited to this, and the number of wheels is changed, and some or all of the wheels are infinite other than wheels. You may comprise so that it may replace with support and drive mechanisms, such as a track | orbit, a leg-shaped body.

Embodiment 4 FIG.
In the fourth embodiment, the robot of the present invention is applied to a variable polyhedron type moving device. In FIG. 17, reference numeral 1 denotes a robot according to Embodiment 4 of the present invention, and the robot 1 is configured by an exoskeleton having an octahedral shape formed by combining eight triangles. Each side of the octahedron is composed of a rod-like body having extendable connecting portions 2, 2,... Driven by direct acting actuators 20, 20,. It is configured as a spherical joint. It is also possible to incorporate a rotary actuator into the spherical joint. Other components such as the drive control device are incorporated in the spherical joints and / or the connecting portions 2, 2,.

  The robot 1 is grounded by disengaging one end of an arbitrary rod-shaped body having the connecting portion 2 from the spherical joint as shown in FIG. 17 (b) from an octahedral shape as shown in FIG. 17 (a). This allows the detached rod-shaped body to become a support shaft and to contact the ground with an arbitrary apex of the octahedron, allowing various irregularities to rest on the ground, and further, moving by extending and retracting the rod-shaped body Is possible.

  The configuration of the robot 1 is substantially the same as the configuration of the first embodiment shown in the block diagram of FIG. 3, and the connecting portions 2, 2,... And the actuators 20, 20,. And a detection mechanism. The drive control apparatus includes a control unit, a recording unit, a storage unit, a measurement unit, an output unit, and an input unit.

  The process performed by the drive control device for the robot 1 according to the fourth embodiment is substantially the same as that of the first embodiment described with reference to FIG. 7, and the controller performs wireless communication as a process corresponding to step S1 in FIG. The target exercise information transmitted at is received by the input means.

Then, as a process corresponding to step S2, the drive control device of the robot 1 derives the target ZMP and the target acting force from the received target motion information by the various calculation methods described in the first embodiment, and proceeds to step S3. As a corresponding process, a force applied to each contact portion is derived so that the derived acting force is optimally distributed based on the norm minimum criterion. In the example shown in FIG. 17, the robot 1 of the present invention derives the floor acting force distributed in the direction of each contact point indicated by the position vectors r _S1 , r _S2 , r _S3 , and r _S4 from the center of gravity CoM.

  Further, as a process corresponding to step S4, the drive control device of the robot 1 calculates a torque value as a control target value of each actuator 20, 20,..., And as a process corresponding to step S5, each calculated control target value. Are output to the respective actuators 20, 20,. By such processing, the robot 1 configured as a variable polyhedron type apparatus performs various motions.

  The first to fourth embodiments are merely a part of the infinite number of realizations of the robot of the present invention, and the robot of the present invention is not limited to the above-described form, and can be developed in various forms. Is possible.

Claims (13)

In a robot having a plurality of drivable connecting parts,
A plurality of actuators for driving the connecting portion based on the torque value received as the control target value;
Calculation means for calculating a torque value of each actuator that drives each connecting portion based on a target acting force set to act on an assumed contact portion with the outside;
Means for outputting the calculated torque value to each actuator as a control target value.   The robot according to claim 1, wherein the actuator is a direct-acting actuator that expands and contracts a connecting portion and / or a rotating actuator that rotates the connecting portion. Means for receiving a setting of the target acting force;
The robot according to claim 1, wherein the calculation unit is configured to calculate a torque value of each actuator based on the received target acting force. Means for receiving target exercise information indicating the target exercise;
A derivation means for deriving a target acting force to be set based on the received target exercise information;
The robot according to claim 1, wherein the calculation unit is configured to calculate a torque value of each actuator based on the derived target acting force.   The robot according to claim 4, wherein the calculating unit is configured to calculate a torque value of each actuator from a target acting force based on a forward kinematic model.   The said calculating means is comprised so that the torque value of each actuator may be calculated in consideration of at least one force in inertia force, Coriolis force, and centrifugal force. robot. It further comprises detection means for detecting the force received from the outside,
The robot according to any one of claims 1 to 6, wherein the calculation unit is configured to calculate a torque value of each actuator in consideration of the force detected by the detection unit. The contact portion is a set of a plurality of contact points,
The said calculating means is comprised so that the torque value of each actuator may be calculated based on the force allocated to each contact point calculated based on the norm minimum norm. The robot according to any one of claims 1 to 7.   9. The method according to claim 1, wherein the calculating means is configured to calculate a torque value of each actuator based on a target acting force so as to compensate for gravity applied to the center of gravity. Crab robot.   10. The torque value of each actuator is calculated based on a restraining force that suppresses internal motion caused by redundant degrees of freedom related to a plurality of connecting portions. The robot according to any one of the above.   The robot according to claim 1, further comprising a plurality of legs that operate by driving the connecting portion. The organism is configured to be wearable,
Means for detecting the force received from the organism,
Means for deriving a target acting force to be set based on the detected force; and
The robot according to claim 1, wherein the calculating unit is configured to record so as to calculate a torque value of each actuator based on the derived target acting force. In a control device for controlling the operation of a robot having a plurality of connecting portions that can be driven by an actuator,
Means for calculating a torque value of each actuator for driving each connecting portion based on a target acting force set to act on an assumed external contact portion;
And a means for outputting the calculated torque value to each actuator as a control target value.

Images