CN114022824A - Narrow environment-oriented quadruped robot motion planning method - Google Patents

Narrow environment-oriented quadruped robot motion planning method Download PDF

Info

Publication number
CN114022824A
CN114022824A CN202111472945.XA CN202111472945A CN114022824A CN 114022824 A CN114022824 A CN 114022824A CN 202111472945 A CN202111472945 A CN 202111472945A CN 114022824 A CN114022824 A CN 114022824A
Authority
CN
China
Prior art keywords
robot
motion
quadruped robot
local
path
Prior art date
Legal status (The legal status 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 status listed.)
Pending
Application number
CN202111472945.XA
Other languages
Chinese (zh)
Inventor
刘勇
张震
晏嘉卿
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
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 Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN202111472945.XA priority Critical patent/CN114022824A/en
Publication of CN114022824A publication Critical patent/CN114022824A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Manipulator (AREA)

Abstract

A motion planning method of a quadruped robot facing a narrow environment comprises the following steps: constructing a local point cloud map according to surrounding environment information detected by an airborne sensor of the quadruped robot, and converting the local point cloud map into a trafficability cost map containing environment topographic information; determining a local target point of the motion planning in the local scope according to the global guide path input by the external global planner; planning a rough optimal collision-free path on the cost map by using an A-x algorithm by using a particle model of the quadruped robot based on the trafficability cost map and the local target points; carrying out track optimization on the motion dynamics model of the quadruped robot by using the rough optimal collision-free path to generate an effective motion track meeting the motion dynamics constraint of the robot; calculating a motion control instruction of the quadruped robot by using a rolling time domain control strategy and utilizing the effective motion track; and judging whether the robot finishes executing the track to reach the target position, if not, returning to the step I, and repeating the whole process until the process is finished.

Description

Narrow environment-oriented quadruped robot motion planning method
Technical Field
The invention belongs to the technical field of robots, and particularly relates to a narrow environment-oriented quadruped robot motion planning method.
Background
Compared with the traditional wheel type robot, the four-foot robot has stronger terrain adaptability and more flexible movement capability, is suitable for running in various complex environments, and can complete challenging tasks such as search and rescue after disasters, interplanetary exploration and the like. Quadruped robots are generally required to operate autonomously in a structured indoor environment with a narrow space and a cluttered obstacle or an unstructured outdoor environment with a rugged terrain and a variety of obstacles. Therefore, the rapid and safe motion trail planning on line has great significance.
At present, a track planning method for a wheeled robot abstracts the robot into a particle model, performs expansion processing on obstacles in the environment, and further plans a collision-free running path by using an algorithm based on graph search or sampling. The whole body of the quadruped robot has 12 degrees of freedom, has very flexible omnidirectional movement capability and strong terrain adaptation capability, has very complex kinematics and dynamics models, and cannot ensure performability and safety and fully exert the movement capability of the quadruped robot by using a movement path planned by an oversimplified description mode.
There have been some studies on the planning of the motion of quadruped robots known so far. For example, the invention patent of chinese patent application No. 201310014887.5 discloses a quadruped robot motion planning method for complex terrain, which plans a fine sequence of foothold points based on the terrain information of the environment to realize passing on the complex terrain. However, due to the dependence on accurate terrain modeling, the real-time performance and robustness are poor. The invention patent of chinese patent application No. 201810827669.6 discloses a method and system for planning a four-legged robot path based on an improved a-x algorithm, which plans a collision-free path in a narrow passage through the improved a-x search algorithm, but only plans a two-dimensional discrete path without considering a special motion dynamics model of the robot, and cannot guarantee the performability and effectiveness of the path.
Therefore, the existing quadruped robot motion planning method still has defects.
Disclosure of Invention
The invention aims to solve the problems and provides a quadruped robot motion planning method oriented to a narrow environment, which considers the motion dynamics characteristics and constraints of the quadruped robot, uses a track optimization technology to plan flexible, robust and reliable motion tracks, and can safely and effectively pass through a narrow and rugged space.
In order to achieve the purpose, the invention adopts the following technical scheme: a motion planning method of a quadruped robot facing a narrow environment comprises the following steps:
the method comprises the following steps that firstly, a local point cloud map is constructed according to surrounding environment information detected by an airborne sensor of the quadruped robot, and the local point cloud map is converted into a trafficability cost map containing environment terrain information;
secondly, determining a local target point which moves and plans in a local range according to a global guide path input by an external global planner;
thirdly, planning a rough shortest collision-free path on the cost map by using an A-shortest path algorithm by utilizing a particle model of the quadruped robot based on the trafficability cost map and the local target point in the local range;
step four, the rough path is used as an initial track, a motion dynamics model of the quadruped robot and obstacle collision constraint are considered for track optimization, and an effective motion track ensuring performability is generated;
step five, calculating a motion control instruction of the quadruped robot by using a rolling time domain control strategy and utilizing the effective motion track;
and step six, judging whether the robot finishes executing the track to reach the target position, if the robot does not finish executing the track to reach the target position, returning to the step one, and repeating the whole process until the process is finished.
Further, the first step specifically comprises:
s11, setting a sliding window to only store point cloud data in a history period of time, and then converting the point cloud data of the environmental barrier acquired by the multi-line laser radar and the stereoscopic vision depth camera into a unified world coordinate system by using three-dimensional pose information output by a robot positioning module and then overlapping the point cloud data;
and S12, converting the local point cloud map formed by splicing the point cloud data of the multi-frame history into a trafficability cost map containing environmental terrain information, wherein the trafficability cost map consists of discrete grids containing trafficability cost values.
Further, the second step is specifically:
s21, setting the length and width range of the local map, and defining the local map range by taking the body of the quadruped robot as the center;
and S22, searching the last track point in the local map range on the global guide path from the current position of the quadruped robot, and using the last track point as a local target point of the local motion plan.
Further, the third step is specifically: based on a grid map containing trafficability cost, searching a trafficable path from the current position of the quadruped robot to a local target point by using a robot particle model through an A-shortest path search algorithm, wherein the path is a collision-free shortest path, and the grid map has fixed spatial resolution, so that the path is regarded as a rough discrete path and has the advantage of extremely high search speed; if the search fails, the local target points are traced backwards gradually along the global path until a path meeting the traffic state is obtained.
Further, the fourth step is specifically:
s41, simplifying and abstracting a motion dynamics model of the quadruped robot, wherein GRFs of support legs of the quadruped robot meet friction cone constraint;
and S42, taking the rough discrete path obtained by searching in the third step as a track initial value, considering the motion dynamics constraint of the quadruped robot and the obstacle collision constraint, and generating a finer effective motion track by using a nonlinear optimization algorithm, wherein the algorithm ensures the performability of the generated track.
Furthermore, in the fourth step, the discrete trajectory is defined to include a series of n robot poses
Figure BDA0003390099360000041
And n-1 time intervals
Figure BDA0003390099360000042
Wherein
Figure BDA0003390099360000043
Representing the two-dimensional coordinates of the robot in a grid map, thetai∈S1Indicating the orientation angle, Δ T, of the robotiRepresenting the slave pose x of the robotiMove to the next pose xi+1The time interval of (c). The motion trajectory is therefore described as a parameterized expression as follows:
Figure BDA0003390099360000044
further, the optimization in step S42 includes optimization of the sum of acceleration and optimization of time of the motion trajectory;
based on the desire for a rapid and flexible movement of the quadruped robot, the shorter the consumption time of the motion trajectory, the better, the objective function for time optimization of the motion trajectory is set to the minimum sum of squares of the time intervals:
Figure BDA0003390099360000045
the optimization of the sum of the accelerations is as follows:
Figure BDA0003390099360000046
further, the step five is specifically as follows: the effective motion trajectory is based on expected speed and acceleration values of the quadruped robot, and only a first group of expected control quantities are transmitted to a gait controller of the quadruped robot as control commands to control the quadruped robot in real time.
Further, since there is a delay in the execution of the control command, a time parameter t for correcting a control error caused by the delay is setd(ii) a The K-th group of desired control quantities on the trajectory is selected as the control command to be finally given to the robot, wherein the value of K is selected in such a way that,
Figure BDA0003390099360000047
ΔTirepresenting the slave pose x of the robotiMove to the next pose xi+1The time interval of (c).
Furthermore, in step 6, if the difference between the current position of the quadruped robot and the position of the local target point is smaller than the set threshold and the moving speed of the robot is zero, the motion planning is ended, and a stop command is sent to the quadruped robot to stop the motion.
Compared with the prior art, the invention has the beneficial effects that: aiming at the motion of the quadruped robot in a narrow environment, the motion planning method is adopted, a properly simplified quadruped robot motion dynamics model is utilized, a nonlinear optimization problem is established, and an optimal motion track is solved, so that the efficiency and the real-time performance of a planning algorithm are ensured, the online rolling time domain control can be realized, the anti-interference performance and the robustness of the robot in the motion process are effectively improved, the generated track can be ensured to meet the motion dynamics characteristics of the quadruped robot, the effectiveness and the track performability of the planning algorithm are ensured, the powerful motion capability of the quadruped robot can be fully exerted, and the narrow and complex environment can be quickly and effectively passed through.
Drawings
FIG. 1 is a schematic diagram of the hardware architecture of a quadruped robot;
FIG. 2 is a schematic diagram of a map of a local point cloud created in the first step;
FIG. 3 is a simplified schematic of a kinematic model of a quadruped robot;
FIG. 4 is a schematic diagram of a trajectory optimization process using a polygonal contour of a quadruped robot for collision detection;
fig. 5 is a trajectory sequence chart generated by the quadruped robot motion planning method of the present invention.
Detailed Description
The present invention will be described in detail with reference to specific examples. The following examples will assist those skilled in the art in further understanding the invention, but are not intended to limit the invention in any way. It should be noted that variations and modifications can be made by persons skilled in the art without departing from the spirit of the invention. All falling within the scope of the present invention.
As shown in fig. 1, a schematic diagram of a quadruped robot structure adopted in this embodiment is shown, and a specific sensor hardware module of the quadruped robot includes: a 16-line rotary laser radar with the model of Velodyne VLP-16, which is horizontally arranged at the front side of the top of the robot, has a vertical field angle of-15 degrees and can detect obstacles within a distance of 0.5-100 m; a stereoscopic vision depth camera, model number Realsense D435i, field angle 79 ° 59 °, mounted 20 ° down from the front of the robot, for detecting ground obstacles within 2m distance; an inertial measurement sensor (IMU) Xsens MTi-300 is used for acquiring the attitude angle and the motion state of the robot body; a synchronization mainboard for providing sensor synchronization signals, and an Intel NUC microcomputer for operating the robot perception and path planning algorithm. The model of the quadruped robot body is Jueying Mini, and 12 joints of the whole body are driven by a motor.
The method for planning the motion of the quadruped robot facing the narrow environment comprises the following specific implementation details:
step 1: the sensors carried by the quadruped robot comprise two types of sensors, namely a laser radar sensor and a depth camera sensor, and the two sensors output point cloud data describing the environment at the current moment at a fixed frame rate (VLP-16 is set to be 10Hz, and D435i is set to be 15 Hz). However, because the sensing range and the visual field of single-frame data captured by the sensor are limited, environmental information around the robot cannot be described in detail, the robot is perfected by using point cloud data of a spliced multi-frame history, as shown in fig. 2.
Specifically, based on three-dimensional position and attitude information provided by an additional robot positioning algorithm module, point clouds captured by a laser radar and a depth camera sensor are respectively converted to a world coordinate system and are overlapped, and then a sliding window is arranged to store point cloud data within 5 seconds of history. Namely, the point cloud data of a new frame at present is added into the sliding window for storage, and the point cloud data acquired 5 seconds before is deleted from the sliding window.
The spliced local point cloud map is converted into a trafficability cost map for describing terrain information, and the trafficability cost map is composed of discrete grids containing trafficability cost values, and the cost values are calculated by the following method:
Figure BDA0003390099360000061
wherein r, s, h respectively represent roughness, slope and height values of the grid, and a threshold rthre、sthre、hthreMaximum value, w, indicating allowable traffic1、w2、w3Represents a weight value of each item, and the sum thereof is 1. In practical use, w can be adjusted according to the field environment1、w2、w3To obtain a traffic cost map that conforms to the robot's motion capabilities.
Step 2: first, the range size of the local map needs to be set according to the actual environment, for example, in a relatively narrow indoor environment, the length and width of the local map may be set to 5 meters × 5 meters, and in an outdoor relatively spacious environment, the length and width may be set to 8 meters. And then selecting a target point of the local planning in the local map range based on the global path of the navigation task, wherein a route point closest to the edge in the local map is searched along the global path from the current position and is used as the target point of the local movement planning.
And step 3: and based on the local trafficability cost map, searching a trafficable path to a local target point from the current position of the robot by using an A-algorithm, and if no trafficable path exists to the local target point, gradually backtracking the target point along the global path in the step 2 until a path meeting the traffic state is obtained.
In order to ensure the efficiency of a path planning algorithm, a simple robot mass point model is adopted in a path searching stage, virtual expansion processing is carried out on the obstacles in the grid map, the expansion radius is set to be the radius of an inscribed circle of the quadruped robot contour projection, and the robot is ensured to pass through a narrow passage area. Traversability information of the terrain is stored as cost values for grids in a cost map. In the path searching stage, the trafficability cost is considered, namely the cost function of the searching algorithm is added to the trafficability cost value, and the algorithm tends to find a path leading to an area with low trafficability cost.
Meanwhile, in the process of path searching, in order to ensure that the generated track is relatively stable every time and avoid changing a sharp path as much as possible, the algorithm keeps the historically planned path and reduces the cost value of the grid where the historical path is located in the searching process, namely punishment is carried out on a new path with large change, so that the algorithm tends to find a path with small deviation from the historical path.
And 4, step 4: the rough optimal collision-free path planned in the step 3 is used as an initial path, the motion path of the quadruped robot is further generated by using an optimization-based method, and the motion dynamics constraint of the robot is taken into consideration in the process, so that the effectiveness and the safety of the motion path are ensured.
As shown in fig. 3, we have extensively modeled the whole body motion dynamics of a quadruped robot. The motion actuators of the robot are 4 bionic legs (identified by LF, RF, LH, RH), each leg consisting of 3 independently driven joint motors, denoted HipX, HipY and Knee respectively. Wherein f is1-4GRFs (ground Reaction forces) is used for expressing the supporting force provided by the ground on the foot of the robot, and in order to ensure the stability and balance of the quadruped robot during movement, the GRF of the supporting leg is required to meet the constraint of the friction cone. We assume that ignoring Coriolis forces while GRF is the only external force acting on the robot, the robot center of mass, CoM (Centre o)f Mass) linear velocity
Figure BDA0003390099360000081
Linear acceleration
Figure BDA0003390099360000082
Angular velocity omegabAngular acceleration
Figure BDA0003390099360000083
The calculation method of (2) is as follows:
Figure BDA0003390099360000084
Figure BDA0003390099360000085
wherein
Figure BDA0003390099360000086
Is the mass of the robot and is,
Figure BDA0003390099360000087
is the vector of the gravity acceleration,
Figure BDA0003390099360000088
is a central moment of inertia and is,
Figure BDA0003390099360000089
is the vector from the center of mass of the robot to the ith foot end, the center of mass position and the rotational inertia I of the robotgObtained by CAD software, pcom,iThe vector is obtained by calculating position feedback data of a robot joint motor. Linear velocity
Figure BDA00033900993600000810
Is decomposed into longitudinal v under the robot coordinate systemxAnd v in the transverse directionyTherefore, the constraint of the robot motion speed is expressed by the following inequality:
vmin≤vx≤vmax
|vy|≤|vbound|,|ω|≤|ωbound|
wherein v isminAnd vmaxSpeed extremes for backward and forward movement, v, respectivelybound、ωboundThe maximum values of the speed of the lateral movement and the steering, respectively, are set according to the actual situation and the operating environment of the robot.
GRFs can be decomposed into 3 components f under the world coordinate systemx、fy、fzThe friction cone constraint is described by the following inequality:
|fz|≤fmax,|fx|≤μ|fz|，|fy|≤μ|fz|
wherein μ is the friction factor of the ground, fmaxThe maximum acting force which can be generated by the leg joint of the robot.
As shown in fig. 4, the polygonal contour of the robot is used for obstacle collision detection in the trajectory optimization process. In the traffic cost map, a grid with a cost value higher than a threshold is considered as an occupied state. The closest distance, denoted d, of the grid of surrounding occupancy states in the environment to the robot profile is then calculatedk(k ═ 1, 2, …, P, where P is the number of obstacle grids), then the obstacle constraint is expressed by the following inequality:
dk≥dmin
wherein d isminAnd adjusting the set minimum distance threshold according to the actual environment and the robot.
In summary, the motion trajectory planning problem of a quadruped robot is defined as a nonlinear optimization problem under the motion dynamics constraint and the obstacle constraint. Defining a discrete trajectory to include a series of n robot poses
Figure BDA0003390099360000091
And n-1 time intervals
Figure BDA0003390099360000092
Wherein
Figure BDA0003390099360000093
Representing the two-dimensional coordinates of the robot in a grid map, thetai∈S1Indicating the orientation angle, Δ T, of the robotiRepresenting the slave pose x of the robotiMove to the next pose xi+1The time interval of (c). The trajectory is therefore described as a parameterized expression as follows:
Figure BDA0003390099360000094
since it is desirable for the robot to move quickly and flexibly, and to some extent to keep the motion smooth and save energy, the objective function of the trajectory optimization problem is set to minimize the sum of the squares of the motion time intervals and the sum of the squares of the accelerations:
Figure BDA0003390099360000095
Figure BDA0003390099360000096
in order to improve the efficiency of the online solver, constraint conditions are added to the objective function as additional secondary penalty terms. The above constraint conditions (obstacle constraint and kinematic dynamics constraint) can be divided into a single boundary constraint and a double boundary constraint, and the approximate penalty function is defined as:
Figure BDA0003390099360000101
Figure BDA0003390099360000102
where γ and e are parameters for adjusting the approximation accuracy of the function. The whole trajectory optimization problem is expressed as a multi-objective nonlinear programming problem:
Figure BDA0003390099360000103
a＝[ΔTi,‖ai‖]T
Figure BDA0003390099360000104
c＝[F2(vi),F2(ωi)]T
wherein the content of the first and second substances,
Figure BDA0003390099360000105
representing the optimal track solution, a, b and c are respectively different optimization target items, W1、W2、W3Is a weight coefficient matrix used for adjusting optimization tendency.
As shown in fig. 5, an optimal trajectory sequence is generated by trajectory optimization, and the polygonal box represents a projection of the robot outline on the ground, so that the trajectory perfectly avoids obstacles in the environment.
And 5: and (4) considering the external environment interference, the local map description and the uncertainty of the dynamic model, and adopting a rolling time domain control strategy to generate a control instruction of the quadruped robot. Based on motion trajectory
Figure BDA0003390099360000106
Calculating a desired velocity and acceleration value of the robot, including a velocity v of the longitudinal movement, according to the following formulaxAcceleration axVelocity v of the transverse movementyAcceleration ayAnd angular velocity ω, angular acceleration α:
Figure BDA0003390099360000107
Figure BDA0003390099360000111
Figure BDA0003390099360000112
Figure BDA0003390099360000113
only the first set of desired control quantities on the trajectory are passed as control commands to the quadruped robot gait controller to effect real-time control of the robot. However, in real-world applications, there is an inevitable time delay in the execution of the local terrain construction and trajectory optimization algorithms. Therefore, we set a time parameter tdFor correcting control errors caused by delays. And selecting the K group of expected control quantities on the track as the control command finally given to the robot, wherein the K value is selected in the following way:
Figure BDA0003390099360000114
step 6: and if the robot reaches the destination, namely the difference between the current position and the given target position is smaller than the set threshold value and the moving speed of the robot is 0, ending the motion planning algorithm and sending an instruction to stop the motion of the quadruped robot.
The above is the preferred embodiment of the present invention, and the scope of the present invention is not limited thereto, and variations and modifications made by those skilled in the art according to the design concept of the present invention should be considered to be within the scope of the present invention. More specifically, various variations and modifications are possible in the component parts and/or arrangements of the subject combination arrangement within the scope of the disclosure, the drawings and the appended claims. In addition to variations and modifications in the component parts and/or arrangements, other uses will also be apparent to those skilled in the art.

Claims (10)

1. A motion planning method of a quadruped robot facing a narrow environment is characterized by comprising the following steps:
the method comprises the following steps that firstly, a local point cloud map is constructed according to surrounding environment information detected by an airborne sensor of the quadruped robot, and the local point cloud map is converted into a trafficability cost map containing environment terrain information;
step two, determining a local target point of the motion planning in a local range according to a global guide path input by an external global planner;
thirdly, planning a rough shortest collision-free path on the cost map by using an A-shortest path algorithm by utilizing a particle model of the quadruped robot based on the trafficability cost map and the local target point in the local range;
step four, the rough path is used as an initial track, a motion dynamics model of the quadruped robot and obstacle collision constraint are considered for track optimization, and an effective motion track ensuring performability is generated;
step five, calculating a motion control instruction of the quadruped robot by using a rolling time domain control strategy and utilizing the effective motion track;
and step six, judging whether the robot finishes executing the track to reach the target position, if the robot does not finish executing the track to reach the target position, returning to the step one, and repeating the whole process until the process is finished.
2. The narrow environment-oriented quadruped robot motion planning method according to claim 1, wherein the first step is specifically as follows:
s11, setting a sliding window to only store point cloud data in a history period of time, and then converting the point cloud data of the environmental barrier acquired by the multi-line laser radar and the stereoscopic vision depth camera into a unified world coordinate system by using three-dimensional pose information output by a robot positioning module and then overlapping the point cloud data;
and S12, converting the local point cloud map formed by splicing the point cloud data of the multi-frame history into a trafficability cost map containing environmental terrain information, wherein the trafficability cost map consists of discrete grids containing trafficability cost values.
3. The narrow environment-oriented quadruped robot motion planning method according to claim 2, wherein the second step is specifically as follows:
s21, setting the length and width range of the local map, and defining the local map range by taking the body of the quadruped robot as the center;
and S22, searching the last track point in the local map range on the global guide path from the current position of the quadruped robot, and using the last track point as a local target point of the local motion plan.
4. The narrow environment-oriented quadruped robot motion planning method according to claim 3, wherein the third step is specifically as follows:
searching a passable path from the current position of the quadruped robot to a local target point by utilizing a robot particle model through an A-shortest path search algorithm based on a grid map containing trafficability cost, wherein the path is a collision-free shortest path, and the path is regarded as a coarse discrete path because the grid map has fixed spatial resolution; if the search fails, the local target points are traced backwards gradually along the global path until a path meeting the traffic state is obtained.
5. The narrow environment-oriented quadruped robot motion planning method according to claim 4, wherein the fourth step is specifically as follows:
s41, simplifying and abstracting a motion dynamics model of the quadruped robot, wherein GRFs of support legs of the quadruped robot meet friction cone constraint;
and S42, taking the rough discrete path obtained by searching in the third step as a track initial value, considering the motion dynamics constraint of the quadruped robot and the obstacle collision constraint, and generating a finer effective motion track by using a nonlinear optimization algorithm, wherein the algorithm ensures the performability of the generated track.
6. The narrow-environment-oriented quadruped robot motion planning method according to claim 5, wherein in the fourth step, the discrete trajectory is defined to include a series of n robot poses
Figure FDA0003390099350000031
And n-1 time intervals
Figure FDA0003390099350000032
Wherein
Figure FDA0003390099350000033
Representing the two-dimensional coordinates of the robot in a grid map, thetai∈S1Indicating the orientation angle, Δ T, of the robotiRepresenting the slave pose x of the robotiMove to the next pose xi+1The motion trajectory is thus described as a parameterized expression as follows:
Figure FDA0003390099350000034
7. the narrow environment-oriented quadruped robot motion planning method according to claim 6, wherein the optimization in step S42 comprises the time optimization and the acceleration sum optimization of the motion trajectory;
based on the desire for a rapid and flexible movement of the quadruped robot, the shorter the consumption time of the motion trajectory, the better, the objective function for time optimization of the motion trajectory is set to the minimum sum of squares of the time intervals:
Figure FDA0003390099350000035
the optimization of the sum of the accelerations is as follows:
Figure FDA0003390099350000036
8. the narrow environment-oriented quadruped robot motion planning method according to claim 6 or 7, wherein the step five is specifically as follows:
the effective motion trajectory is based on expected speed and acceleration values of the quadruped robot, and only a first group of expected control quantities are transmitted to a gait controller of the quadruped robot as control commands to control the quadruped robot in real time.
9. The narrow-environment-oriented quadruped robot motion planning method according to claim 8, wherein a time parameter t for correcting a control error caused by a delay is set due to a delay in the execution of the control commandd(ii) a The K-th group of desired control quantities on the trajectory is selected as the control command to be finally given to the robot, wherein the value of K is selected in such a way that,
Figure FDA0003390099350000041
ΔTirepresenting the slave pose x of the robotiMove to the next pose xi+1The time interval of (c).
10. The narrow-environment-oriented quadruped robot motion planning method according to claim 1, wherein in step 6, if the difference between the current position of the quadruped robot and the position of the local target point is less than a preset threshold and the moving speed of the robot is zero, the motion planning is ended, and a stop command is sent to the quadruped robot to stop the motion.
CN202111472945.XA 2021-12-03 2021-12-03 Narrow environment-oriented quadruped robot motion planning method Pending CN114022824A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111472945.XA CN114022824A (en) 2021-12-03 2021-12-03 Narrow environment-oriented quadruped robot motion planning method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111472945.XA CN114022824A (en) 2021-12-03 2021-12-03 Narrow environment-oriented quadruped robot motion planning method

Publications (1)

Publication Number Publication Date
CN114022824A true CN114022824A (en) 2022-02-08

Family

ID=80067869

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111472945.XA Pending CN114022824A (en) 2021-12-03 2021-12-03 Narrow environment-oriented quadruped robot motion planning method

Country Status (1)

Country Link
CN (1) CN114022824A (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114625129A (en) * 2022-02-22 2022-06-14 中国科学院自动化研究所 Motion control method and system of position-control leg and foot robot
CN114639088A (en) * 2022-03-23 2022-06-17 姜妹英 Big data automatic navigation method
CN115047875A (en) * 2022-06-07 2022-09-13 中国北方车辆研究所 Task-driven coarse-grained iterative model of quadruped robot
CN115061470A (en) * 2022-06-30 2022-09-16 哈尔滨理工大学 Unmanned vehicle improved TEB navigation method suitable for narrow space
CN115143964A (en) * 2022-07-05 2022-10-04 中国科学技术大学 Four-footed robot autonomous navigation method based on 2.5D cost map
CN115951672A (en) * 2022-12-15 2023-04-11 锐趣科技(北京)有限公司 Method for robot to pass through narrow passage
CN116578088A (en) * 2023-05-04 2023-08-11 浙江大学 Outdoor autonomous mobile robot continuous track generation method, system, medium and equipment

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114625129A (en) * 2022-02-22 2022-06-14 中国科学院自动化研究所 Motion control method and system of position-control leg and foot robot
CN114625129B (en) * 2022-02-22 2023-09-12 中国科学院自动化研究所 Motion control method and system of position control leg-foot robot
CN114639088A (en) * 2022-03-23 2022-06-17 姜妹英 Big data automatic navigation method
CN115047875A (en) * 2022-06-07 2022-09-13 中国北方车辆研究所 Task-driven coarse-grained iterative model of quadruped robot
CN115047875B (en) * 2022-06-07 2024-05-14 中国北方车辆研究所 Task-driven four-foot robot coarse granularity iteration model
CN115061470A (en) * 2022-06-30 2022-09-16 哈尔滨理工大学 Unmanned vehicle improved TEB navigation method suitable for narrow space
CN115143964A (en) * 2022-07-05 2022-10-04 中国科学技术大学 Four-footed robot autonomous navigation method based on 2.5D cost map
CN115143964B (en) * 2022-07-05 2024-05-10 中国科学技术大学 Four-foot robot autonomous navigation method based on 2.5D cost map
CN115951672A (en) * 2022-12-15 2023-04-11 锐趣科技(北京)有限公司 Method for robot to pass through narrow passage
CN115951672B (en) * 2022-12-15 2024-04-12 锐趣科技(北京)有限公司 Method for robot to pass through narrow channel
CN116578088A (en) * 2023-05-04 2023-08-11 浙江大学 Outdoor autonomous mobile robot continuous track generation method, system, medium and equipment
CN116578088B (en) * 2023-05-04 2024-02-09 浙江大学 Outdoor autonomous mobile robot continuous track generation method and system

Similar Documents

Publication Publication Date Title
CN114022824A (en) Narrow environment-oriented quadruped robot motion planning method
Kim et al. Vision aided dynamic exploration of unstructured terrain with a small-scale quadruped robot
Monteiro et al. Attractor dynamics approach to formation control: theory and application
Song et al. Navigation control design of a mobile robot by integrating obstacle avoidance and LiDAR SLAM
Ren et al. A new fuzzy intelligent obstacle avoidance control strategy for wheeled mobile robot
Siegwart et al. Autonomous mobile robots
González-Sieira et al. Autonomous navigation for UAVs managing motion and sensing uncertainty
Zhang et al. Efficient motion planning based on kinodynamic model for quadruped robots following persons in confined spaces
Sekhavat et al. Motion planning and control for Hilare pulling a trailer: experimental issues
CN114442621A (en) Autonomous exploration and mapping system based on quadruped robot
US20230278214A1 (en) Robot localization using variance sampling
Rioux et al. Autonomous SLAM based humanoid navigation in a cluttered environment while transporting a heavy load
Ollero et al. Control and perception components for autonomous vehicle guidance. Application to the ROMEO vehicles
Balaram Kinematic state estimation for a Mars rover
US9395726B1 (en) Methods and devices for bound and gallop gaits
Zhao et al. A real-time low-computation cost human-following framework in outdoor environment for legged robots
Chen et al. Feedback control for autonomous riding of hovershoes by a cassie bipedal robot
Liu et al. Research on path planning of quadruped robot based on globally mapping localization
Kim et al. Single 2D lidar based follow-me of mobile robot on hilly terrains
Stasse et al. Integrating walking and vision to increase humanoid autonomy
Wang et al. Obstacle detection and obstacle-surmounting planning for a wheel-legged robot based on Lidar
Brumitt et al. Dynamic trajectory planning for a cross-country navigator
AU2021448614A9 (en) Precise stopping system and method for multi-axis flatbed vehicle
AU2021448614A1 (en) Precise stopping system and method for multi-axis flatbed vehicle
He et al. Research on mobile robot positioning and navigation system based on multi-sensor fusion

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination