Locally-optimal Path Planning by Using Probabilistic Roadmaps and Simulated Annealing

It is presented an algorithm for locally-optimal path planning which is based on Kavraki and Latombe’s Probabilistic Roadmaps (PRM) and Simulated Annealing (SA). The PRM method is able to generate collision-free paths in a very short time (once that the roadmap has been constructed). However, when generating the paths, it is not taken into consideration any optimization criteria. It means that it would be interesting trying to reduce the cost of the paths. We propose in this work a method based on a simulated annealing to solve this optimization problem. SA is a method that has been successfully applied in complex optimization problems, and it seems straightforward to think of its application for optimizing a path provided by PRM. In other words, PRM are used to generate one valid solution, and SA is used to optimize it. The results show that the method can be applied for off-line planning (the SA took between 15 and 30 seconds in most of the cases), and we established a set of results that can be used in order to compare the performance of SA against other optimization algorithms.