Neural dynamics based approaches to exploration and tracking control of mobile robots

Neural networks have been used to solve many different problems across several fields of research. In this thesis a neural dynamics approach is applied to three problems in the field of mobile robotics: motion control, multiple target path planning and exploration. For motion control, the proposed controller is an improvement of an existing tracking controller that uses a gated dipole model. To increase its effectiveness an extended Kalman filter position tracking module is added. For multiple target path planning a shunting model neural network is used. The algorithm is capable of generating an efficient path for static, prioritized and dynamic environments. The exploration algorithm uses an occupancy grid map as input to a shunting model neural network. The algorithm attracts the mobile robot to unexplored areas while at the same time encouraging the robot to maintain its current heading and to maintain a safe distance from known obstacles.