Route planning method for mobile robot

Abstract

A route planning method for a mobile robot comprises the steps that the distance between the mobile robot and an obstacle is detected through a plurality of distance sensors; a fish swarm algorithm with an elimination mechanism is provided; the mobile robot is controlled by a controller with a neural network; the neural network updates a neural network weight through the fish swarm algorithm; the neural network is trained through a known map template; the walking route of the mobile robot in environmental space with unknown environment information is planned through the trained neural network. The controller with the neural network structure controls the mobile robot through the fish swarm algorithm with the elimination mechanism added, the robot only needs to be trained in one template map, the behaviors of obstacle avoiding and destination reaching can be learned through the generalization performance of the neural network, a generalized behavior is learned through the algorithm rather than a route track of a specific map, and the adaptive capacity of the robot for complex position environment is improved.

Description

technical field [0001] The invention belongs to the technical field of path planning, in particular to a path planning method for a mobile robot. Background technique [0002] The path planning of a mobile robot means that the robot perceives its environment based on the information obtained by the sensor camera and autonomously plans a route to reach the target state. [0003] Traditional path planning methods for mobile robots have the following problems: [0004] Document 1 (Song Yong, Li Yibin, Li Caihong. Initialization of Reinforcement Learning for Mobile Robot Path Planning[J]. Control Theory and Application, 2012, 29(012):1623-1628) Aiming at the characteristics of slow convergence of existing path planning algorithms, proposed A Q-learning initialization algorithm based on artificial potential field is proposed, and the prior knowledge of the environment is obtained according to the artificial potential field, so that the Q value in the algorithm is initialized to ...

AI Technical Summary

Problems solved by technology

[0010] The artificial fish swarm algorithm is developed by the literature (Li Xiaolei, Shao Zhijiang, Qian Jixin. An optimal model based on animal autonomous bodies: fish swarm algorithm [J]. System Engineering Theory and Practice, 2002, 22(11): 32-38) A multi-point heuristic random search algorithm is proposed. The algorithm is based on the natural behavior of fish foraging in nature, which makes the algorithm have excellent optimization ability, but the algorithm has low versatility and needs to be redone for each new map. train

Images