Abstract
Recent studies of obstacle avoidance under the Dynamic Movement Primitives (DMPs) framework are limited on the dotted obstacles. A novel obstacle avoidance algorithm based on DMPs is proposed for shaped obstacle avoidance. By combining the steering behavior method and the potential field method with the evaluation of Euclidean distance, while absorbing the switching strategy, this new algorithm performs well in mimicking the learned DMPs trajectory and handles the fluctuations occur in the initial stage of motion. The simulations of 2D DMPs trajectory learning of one-obstacle avoidance and multi-obstacle avoidance show that the algorithm is feasible with figurate obstacles, and acquires great conformability with the learned trajectory and smoothness.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
