Mobile Robots Path Planning Based On Dynamic Movement Primitives

Traditional robot path planning methods are usually based on environment modeling which are added lots of constraint conditions.The feasible path is searched through different algorithms,thus the results of path depends on the accuracy of the environment modeling and effectiveness of the algorithm.A novel robot path planning method is proposed from the perspective of learning;it means that the robot is controlled to move to the goal position by the people,and then the robot could move to the target location autonomously.The main research work is as follows:Firstly,the development process and status quo of mobile robot are summarized;dynamic movement primitives is applied to the mobile robot path planning after home and abroad commonly applied research about DMPs are analyzed.It is motivated by the idea of learning from demonstration,the study of this article is put forward.Secondly,the basic theory of dynamic movement primitives is expounded,the application about this algorithm of learning on single DOF and Multi-DOF are introduced.On this basis,the ability of generalization about DMPs is proposed.Thirdly,the platform is built to collect a set of sample data;in the environment with obstacles,the avoidance path can be produced through adding coupling factor.Based on the library of movement primitives,generalization comes true to new goals while the goal is changed.Finally,the simulation and experimental results show that DMPs algorithm is feasible on a mobile robot path planning.The performance about veracity,generalization and adaptability are discussed and compared by a lot of simulations and experiments.The results demonstrate the effectiveness of the proposed algorithm.