Robot demonstrating track generating method based on time-space characteristics

Abstract

The invention relates to a robot demonstrating track generating method based on time-space characteristics. The robot demonstrating track generating method includes steps of 1) recording demonstrating tracks of various degrees of freedom at the tail end of a robot at the same sampling time interval by operating the robot and demonstrating the like tracks multiple times; 2) extracting the common time-space characteristics from the various same types to generate a track template; 3) setting up a target equation meeting the time-space characteristics, and solving the motion track of the robot by a covariant gradient descent method; 4) utilizing the motion track as the reference track to input into an executing mechanism of the robot, tracing the tracks by a controller of the robot so as to enable the generated tracks to be consistent with the reference track. Compared with the prior art, the robot demonstrating track generating method is added with the time-space characteristics obtained by learning of the demonstrating tracks on the basis of complying with original constraints; by adding the constraints, search space of the tracks can be limited and solution speed is accelerated.

Description

technical field [0001] The invention relates to a trajectory generation technology applied to an unmanned vehicle and a multi-joint robot, in particular to a method for generating a robot teaching trajectory based on time-space characteristics. Background technique [0002] Robot trajectory planning, especially for high-dimensional multi-joint robots, has always been the focus and difficulty in the field of robotics. Traditional robot trajectory planning methods can be divided into sampling-based and optimization-based trajectory planning methods. These two Both methods need to define a suitable objective function to optimize the generated trajectory. Therefore, the objective function actually determines the feasibility of generating trajectories. The objective function generally includes initial and target states, dynamics and kinematics models, obstacle information, etc. [0003] Teach-and-learn, as an anthropomorphic trajectory generation method, expects to generate a d...

AI Technical Summary

Problems solved by technology

Due to the limitations of the algorithm itself, many algorithms can only be used for a specific occasion, and often require a small difference in time and space from the taught trajectory, otherwise the desired tra

Images