CN-121979199-A - Robot trajectory control method and device, robot and computer program product

CN121979199ACN 121979199 ACN121979199 ACN 121979199ACN-121979199-A

Abstract

The application discloses a robot track control method, a robot track control device, a robot and a computer program product. The method comprises the steps of obtaining at least one group of teaching data, wherein each group of teaching data corresponds to different teaching tracks, calculating a probability model of a weight vector and a probability model of phase speed based on each group of teaching data, and constructing a factor graph based on the probability model of the weight vector, the probability model of the phase speed, at least one preset waypoint, a nominal phase value corresponding to each waypoint and an accuracy parameter, wherein the factor graph comprises a phase factor, a phase speed factor, a weight vector factor and a waypoint factor, and calculating a task track of the target task based on the factor graph so as to control the robot to execute the target task according to the motion track. By the scheme of the application, the generated track can simultaneously meet a plurality of task waypoints.

Inventors

WU JIAJUN
YAN MENG
CHEN SHENHAO
QIN WENLONG
ZHENG YU

Assignees

深圳市优必选科技股份有限公司

Dates

Publication Date: 20260505
Application Date: 20251223

Claims (10)

1. A robot trajectory control method, comprising: Acquiring at least one group of teaching data, wherein each group of teaching data corresponds to different teaching tracks respectively; Based on the teaching data of each group, calculating a probability model of the weight vector and a probability model of the phase velocity: aiming at a specified target task, constructing a factor graph based on a probability model of the weight vector, a probability model of the phase speed, at least one preset waypoint, a nominal phase value corresponding to each waypoint and a precision parameter, wherein the factor graph comprises a phase factor, a phase speed factor, a weight vector factor and a waypoint factor; And solving a task track of the target task based on the factor graph so as to control the robot to execute the target task according to the motion track.
2. The robot trajectory control method according to claim 1, wherein the constructing a factor graph based on the probability model of the weight vector, the probability model of the phase velocity, the preset at least one waypoint, the nominal phase value and the precision parameter corresponding to each of the waypoints, the factor graph including a phase factor, a phase velocity factor, a weight vector factor, and a waypoint factor includes: constructing phase factors in the factor graph by taking nominal phase values corresponding to the road points as constraints; constructing a phase velocity factor in the factor graph by taking the probability model of the phase velocity as constraint; constructing weight vector factors in the factor graph by taking the probability model of the weight vector as constraint; And constructing a waypoint factor in the factor graph by taking each waypoint and the corresponding precision parameter as constraint.
3. The robot trajectory control method according to claim 1, wherein the calculating the task trajectory of the target task based on the factor graph includes: converting the factor graph to a least squares problem; And solving the least square problem, and generating a task track of the target task based on a solving result.
4. The robot trajectory control method according to claim 3, wherein the task trajectory of the target task is generated by solving the least squares problem based on a result of the solving, and the method includes: determining gradient information of the waypoint factors based on a gradient calculation formula of the waypoint factors; Solving the least square problem by combining the gradient information to obtain an optimal solution of the weight vector, and dividing the optimal solution in a preset phase value range based on a specified step number to obtain a phase value sequence; And calculating to obtain a task track based on the optimal solution of the weight vector, the phase value sequence and a preset probability motion primitive formula.
5. The method of claim 4, wherein the calculating a task trajectory based on the optimal solution of the weight vector, the phase value sequence, and a preset probabilistic motion primitive formula comprises: Traversing the phase value sequence to determine a current phase value; Calculating a task track point corresponding to the current phase value based on the optimal solution of the weight vector, the current phase value and the probability motion primitive formula; And returning to the step of traversing the phase value sequence and the subsequent step of determining the current phase value until the phase value sequence is traversed.
6. The robot trajectory control method of claim 1, wherein the acquiring at least one set of teaching data includes: Controlling the robot to move according to each preset teaching track based on a specified function so as to acquire a group of original teaching data corresponding to each teaching track in each moving process of the robot; And carrying out normalization processing on each group of original teaching data to obtain at least one group of teaching data.
7. A robot trajectory control device, comprising: the acquisition module is used for acquiring at least one group of teaching data, wherein each group of teaching data corresponds to different teaching tracks respectively; The calculation module is used for calculating a probability model of the weight vector and a probability model of the phase speed based on the teaching data of each group: The construction module is used for constructing a factor graph based on the probability model of the weight vector, the probability model of the phase speed, at least one preset waypoint, a nominal phase value corresponding to each waypoint and an accuracy parameter aiming at a specified target task, wherein the factor graph comprises a phase factor, a phase speed factor, a weight vector factor and a waypoint factor; and the control module is used for solving the task track of the target task based on the factor graph so as to control the robot to execute the target task according to the motion track.
8. The robot trajectory control device of claim 7, wherein the build module includes: the first construction submodule is used for constructing phase factors in the factor graph by taking nominal phase values corresponding to the road points as constraints; The second construction submodule is used for constructing the phase velocity factors in the factor graph by taking the probability model of the phase velocity as constraint; A third construction submodule, configured to construct weight vector factors in the factor graph with a probability model of the weight vector as a constraint; and a fourth construction submodule, configured to construct a waypoint factor in the factor graph with each waypoint and the corresponding precision parameter as constraints.
9. A robot comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the method according to any of claims 1 to 6 when executing the computer program.
10. A computer program product, characterized in that it comprises a computer program which, when executed by one or more processors, implements the method according to any one of claims 1 to 6.

Description

Robot trajectory control method and device, robot and computer program product Technical Field The application belongs to the technical field of robots, and particularly relates to a robot track control method, a robot track control device, a robot and a computer program product. Background In the field of imitation learning, motion primitives are often used to let a robot learn task actions through a small number of teaching trajectories and generate motion trajectories that meet task requirements based on these teaching trajectories. The probability motion primitive method can generate a posterior track meeting requirements according to a given task target on the basis of describing the statistical characteristics of the teaching track, so that the probability motion primitive method has higher application value in robot motion planning. However, in existing probabilistic motion primitive methods, trajectory generation typically relies on constraints of a single task object. For example, when a task requires a robot to pass a particular waypoint in a trajectory, the method can generate a motion trajectory that satisfies the waypoint constraints. However, when a task includes multiple target constraints or multiple key location points, it is difficult for the existing probabilistic motion primitive method to simultaneously meet these waypoint requirements, and the posterior trajectory solution process usually only can process a single task target, resulting in limited adaptability and flexibility of generating a trajectory. Based on the above, how to enable the robot to simultaneously meet a plurality of task waypoints on the premise of keeping the statistical characteristics of the teaching track, and to realize the generation of the motion track required by the complex task, becomes a technical problem to be solved in the current urgent need. Disclosure of Invention The application provides a robot track control method, a robot track control device, a robot and a computer program product, which can enable a generated track to simultaneously meet a plurality of task waypoints. In a first aspect, the present application provides a robot trajectory control method, including: acquiring at least one group of teaching data, wherein each group of teaching data corresponds to different teaching tracks respectively; Based on each set of teaching data, a probability model of the weight vector and a probability model of the phase velocity are calculated: aiming at a designated target task, constructing a factor graph based on a probability model of a weight vector, a probability model of a phase speed, at least one preset waypoint, a nominal phase value corresponding to each waypoint and an accuracy parameter, wherein the factor graph comprises a phase factor, a phase speed factor, a weight vector factor and a waypoint factor; And calculating a task track of the target task based on the factor graph so as to control the robot to execute the target task according to the motion track. In a second aspect, the present application provides a robot trajectory control device, comprising: The acquisition module is used for acquiring at least one group of teaching data, wherein each group of teaching data corresponds to different teaching tracks respectively; the calculation module is used for calculating a probability model of the weight vector and a probability model of the phase speed based on each group of teaching data: the construction module is used for constructing a factor graph based on a probability model of a weight vector, a probability model of a phase speed, at least one preset waypoint, a nominal phase value corresponding to each waypoint and a precision parameter aiming at a specified target task, wherein the factor graph comprises a phase factor, a phase speed factor, a weight vector factor and a waypoint factor; And the control module is used for solving the task track of the target task based on the factor graph so as to control the robot to execute the target task according to the motion track. In a third aspect, the present application provides a robot comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the method according to the first aspect when executing the computer program. In a fourth aspect, the present application provides a computer readable storage medium storing a computer program which, when executed by a processor, performs the steps of the method of the first aspect. In a fifth aspect, the present application provides a computer program product comprising a computer program which, when executed by one or more processors, implements the steps of the method of the first aspect described above. Compared with the prior art, the method has the beneficial effects that the scheme realizes unified expression of the characteristics of the multiple teaching tracks based on the probability modeling and