An Efficient Hopfield-type Neural Network Technique for Real-time Vehicle Navigation

Cheng,Siqi, Bo Cui, Xingzhong Zhang, Chaomin Luo, Mohan Krishnan, and Mark Paulik

Path planning is an essential issue for intelligent vehicles and many other robotic applications.  Real-time path planning is desirable for efficient performance in many applications. In this project, a novel Hopfield-type neural-network-based model is proposed for real-time path planning of autonomous vehicles in a known environment. The proposed model is compared with an existing neural networks path planning method. The proposed method does not need any templates, even in unknown environments.  A local map composed of square cells is created through the neural dynamics during the path planning with limited sensory information. A PID controller algorithm for wall-following is also implemented in a mobile robot.

Simulation studies of the proposed approach with the neural networks path planning approach show that the proposed method is capable of planning more reasonable and shorter collision-free paths. The Hopfield-type neural network model is successfully implemented for the vehicle navigation on an actual mobile robot.