| Most robotic systems treat robot motion planning as one of the most important research areas. A motion planner that determines a collision-free path for the robot to achieve a task requires an efficient and flexible method.; A method that extends the neural dynamics framework to include the nonholonomic constraint for a car-like robot is developed in this thesis. In the proposed approach, the robot behaviour such as target acquisition and obstacle avoidance are completely controlled by two control variables, the heading direction and the forward velocity of the robot. The dynamics of these variables, which generates the collision-free path and velocity control commands of the robot, is characterized by a biologically inspired shunting model, whose inputs are modeled from the target and obstacles that are acquired relying on measurable sensor information only. The target input produces an attractive contribution, while the obstacle inputs form repulsive contributions to the robot. Each contribution votes for a certain value of control variables that have unique values at a certain time. The nonholonomic constraint of the mobile robot is fully respected; hence, the resulting path is not just feasible but also easy to control.; A series of computation simulations of a mobile robot in various work environments are demonstrated. The proposed approach is also implemented by a novel tracking controller, which considers the both dynamic and kinematic constraints of a mobile robot. All performances show that the proposed motion planner is structurally effective, and practical for a real robotic system. |
