Local path planning method for intelligent unmanned system based on dual back propagation neural network

Abstract

The invention relates to a local path planning method for an intelligent unmanned system based on a dual back propagation neural network. For the problem of dynamic obstacle avoidance of the intelligent unmanned system, firstly, an environment model of the operation of the intelligent unmanned system is established; secondly, a kinematics model of the operation of the unmanned system is established; thirdly, the dual back-propagation neural network with a topological structure formed by two back propagation neural networks is designed, the local path planning is carried out, a first back-propagation neural network is properly trained offline, the first back-propagation neural network plays a leading role in the dynamic obstacle avoidance, a second first back-propagation neural network is trained offline to obtain output compensation. Finally, the method can be used for an intelligent unmanned system such as a UAV (unmanned aerial vehicle), an unmanned vehicles and an unmanned ship to avoid static obstacles and dynamic obstacles, for obstacles with constantly changing speeds, the method can perform prediction and avoidance according to previous motion of the obstacles, the number ofdecisions and computational burden of the intelligent unmanned system are reduced, and the collision avoidance performance and efficiency are improved.

Description

technical field [0001] The invention relates to an intelligent local path planning method for an unmanned system based on a double backpropagation neural network, which can be applied to intelligent unmanned systems such as unmanned aerial vehicles, unmanned vehicles, and unmanned ships for local path planning to avoid static obstacles and dynamic obstacles. Background technique [0002] With the rapid development of artificial intelligence, the use of intelligent unmanned systems to complete specific tasks has attracted more and more attention from the industry. Path planning is divided into global path planning and local path planning. The global path planning, that is, the environment information is completely known to the unmanned system. Local path planning means that in a dynamic environment with unknown obstacles, the intelligent unmanned system uses its own sensor system to perform fusion processing based on the real-time perceived environmental information and cha...

AI Technical Summary

Problems solved by technology

[0009] The present invention aims at the shortcomings in the above local path planning method that the unmanned system cannot perform continuous and accurate obstacle avoidance, too many times of planning, and cannot obtain the global optimal solution, and provides a new double backpropagation neural network (backpropagation neural network) ) topology, which improves the obstacle avoidance efficiency of the mobile unmanned system and reduces the computational pressure

Images