Reflexive Autonomous Vehicle Control Using Neural Networks

신경회로망을 이용한 반사적인 무인차 제어

In this paper, we have shown a new approach of neural networks for mobile robot motion control under an indoor refracted environment. The vehicle has two powered wheels and four passive casters which support a free motion. And it also uses sonar sensors, infrared sensors, Internal odometer, and contact sensors. Two experiments were conducted to demonstrate our objectives. The first one is that the vehicle executes a reflexive motor control to maintain a constant distance to the boundary. The second one is that as well as the boundary following, the vehicle makes a block obstacle avoidance during its path. Without prior knowledge of external environment. we have accomplished the tasks by employing a simple, reactive stimulus-response neural network scheme associating sensor data with the vehicle's action.