Research On Autonomous Humanoid Robot Path Planning

Robot path planning is the core issue in the research of robot navigation. Not only the results of path planning affect the motion accuracy and timeliness of robot, but also the operation efficiency and stability of programming algorithm affect the efficiency of the robot. This thesis studies global path planning for static environments and path planning problem in the dynamic environment, and then combine them for path planning. The main researches are as follows:Firstly, research on global path planning problem for autonomous robot in the static environment. We use grid method for environmental modeling. In order to improve the convergence speed, an improved nonlinear dynamic inertia weight particle swarm algorithm is proposed. Taking fully account of the multi-objective constrained in path planning, length and smoothness of the path were added to the fitness function. Comparing the PSO and improved PSO in simulation experiments, we can make a conclusion that in both complex environment, the improved algorithm can quickly find the optimum path from the origin to the destination, which path length and the convergence speed and smoothness of path were comparatively well.Secondly, aiming at real-time path planning issue in dynamic environment, the integration of fuzzy control and the improved potential field algorithm is proposed for path planning. By analyzing the shortcomings of artificial potential field method, we add the relative speed and the distance between the target and the robot to artificial potential field model. Considering the speed and direction of obstacles making effect on robot path, fuzzy control algorithm is fused to achieve real-time adjustment of the repulsion function parameters. So that the robot quickly avoid the dynamic obstacles. Simulation results show that the fusion of two algorithms can plan out a real-time, fast and optimum path in a dynamic environment.Finally, using NAO humanoid robot as experimental platform, we establish the world coordinate system and the local coordinate system, and then use the algorithm of Chapter Ⅲ for global planning then extract local target point. During walking process, NAO robot use ultrasonic sensors for real-time detection of environment. When an obstacle is detected, the robot makes adjustments depended on dynamic algorithm described in Chapter IV, so that the robot avoids dynamic obstacles while taking the global optimal paths into account. Experiments show that the feasibility and superiority of algorithm. Therefore, the path planning method combining the two algorithms proposed in this paper can plan a real-time and fast and strong optimum path.