Neural network application in robot motion planning

Abstract

The application of neural networks to real-time motion planning of robotic systems is studied. The proposed framework, using biologically inspired neural networks, for robot motion planning with obstacle avoidance in a nonstationary environment is computationally efficient. The neural dynamics of each neuron in the topologically organized neural network is characterized by a simple shunting equation derived from Hodgkin and Huxley's (1952) membrane model. The real-time optimal robot motion is planned through the dynamic activity landscape of the neural network that represents the dynamic environment. The proposed model can deal with point mobile robots, manipulation robots, holonomic and nonholonomic car-like robots and multi-robot systems. The efficiency and effectiveness are demonstrated by simulation studies.