Trajectory optimization method for spraying robot based on reinforcement learning

Abstract

The invention relates to a trajectory optimization method for a spraying robot based on reinforcement learning. The trajectory optimization method determines the optimal spraying trajectory through image acquisition, image processing, graphic three-dimensional reconstruction, graphic discretization, initial point selection and optimal trajectory strategy selection, wherein the image processing includes camera calibration, image correction and stereo matching. The advantages of the invention lie in that the trajectory optimization method for the spraying robot based on reinforcement learning can plan a feasible path according to different states in different environments, is short in decision-making time and high in success rate and can meet the real-time requirements of online planning, thereby overcoming the shortcomings of poor real-time performance and great calculation amount of the traditional manipulator path planning method.

Description

technical field [0001] The invention belongs to the field of intelligent algorithm control, in particular to a trajectory optimization method for a spraying robot based on reinforcement learning. Background technique [0002] Most of the spraying robots use teaching spraying, and spraying according to the trajectory preset by the technical staff. This spraying method is effective in spraying according to the previous spraying experience, but the early programming work is relatively large, and the trajectory optimization is not obvious. Efficiency Low, the paint waste is relatively large. [0003] Trajectory optimization based on reinforcement learning uses reinforcement learning to model the environment through robots, and simulates the same or similar conditions as the environment inside the machine to optimize the spraying trajectory. Among the existing trajectory optimization algorithms, they are usually probabilistic road map method, fast search tree method and artifici...

AI Technical Summary

Problems solved by technology

When this type of method will model the environment or simulate the space, it needs to sample various postures of the manipulator, and judge whether the current action is reasonable through kinematic equations, and the amount of calculation is large; when obstacles and target positions in the environment occur When changing, it is necessary to recalculate the mapping of the configuration space for the new environment, and it is difficult to achieve the purpose of dynamic real-time planning

In addition, such methods need to build an accurate physical model of the manipulator in advance, and the deviation of the modeling will directly affect the effect of the manipulator control

Images