Time Decaying Mapping for Reactive Obstacle Avoidance


In this project, we develop a method to avoid the local minima problem faced by reactive obstacle avoidance algorithms.


One of the basic issues in navigation of mobile robots is the obstacle avoidance task which is commonly achieved using reactive control paradigm where a local mapping from perceived states to actions is acquired. The algorithms of this class suffer from a major drawback of exhibiting cyclic behavior when encountered with certain obstacle configurations. This paper presents a cognitive time decaying approach to overcome this cyclic behavior .The Dynamic Window algorithm is taken as an example for implementing this approach. To build a dynamic window based obstacle avoider, we use time decaying heuristic function for history mapping - which innately eliminates local minima even for a cluttered environment and gives the robot an exploratory nature best suited for map building purposes. The algorithm is successfully tested on a simulation, where it is shown to avoid the U bend problem of local minima.


S. Arora, S.Indu, "A Novel Time Decaying Approach to Obstacle Avoidance." Proc. International Conference on Pattern Recognition and Machine Intelligence, Delhi, India, 2009. (pdf)