| Navigation of intelligent mobile robots is a broad topic that covering a large spectrum of different technologies, and it is vitally important to many applications. At present, autonomous navigation, especially in the unknown and dynamic environments, has been a hot issue. Its aim is to move purposely and accomplish the task without any guidance. This requires the robot can obtain a safe feasible and optimal trajectory from the robot's current position to the goal position with the capacities of self-organizing, self-planning and self-adapting well in the complex environments.This paper proposes a novel improved method called coordinating potential field (CPF), in which the functions of repulsion and coordinating vector are newly designed. The shortcomings during robot navigation in dynamic environment that using the traditional artificial potential field (APF) method are solved successfully. The local potential field is constructed by using local subgoals, which can be obtained by updating dynamic windows. Then this paper puts forward a multi-objective path-selecting algorithm by which the robot can select an optimal strategy. The algorithm makes emphasis on the dynamic obstacle avoidance.In order to improve the effect of whole course optimization, a multi-objective parameter optimization using adaptive genetic algorithm and a practicable fitness function are implemented. The algorithm can guide the robot to plan a safe, smooth and short path successfully. In this section, we lay emphases on analyzing how to avoid the dynamic obstacles. Simulation results indicate that this strategy is practicable and effective.Then we design the motion control for the nonholonomic wheeled robots system by considering dynamics and kinematics constraints. The cubic spiral curve was used to smooth the collision-free path that the robot can track easily. Through controlling the velocity of two driven wheels, point-to-point stabilization and path tracking can be realized. |
PDF Full Text Request