Predicting the Performance of Randomized Parallel Search An Application to Robot Motion Pla.pdfVIP

Journal of Intelligent and Robotic Systems 38: 31–53, 2003. 31 © 2003 Kluwer Academic Publishers. Printed in the Netherlands. Predicting the Performance of Randomized Parallel Search: An Application to Robot Motion Planning DANIEL J. CHALLOU, MARIA GINI, VIPIN KUMAR and GEORGE KARYPIS Department of Computer Science and Engineering, University of Minnesota, Minneapolis, MN 55455, U.S.A.; e-mail: gini@cs.umn.edu Abstract. In this paper we discuss methods for predicting the performance of any formulation of randomized parallel search, and propose a new performance prediction method that is based on obtaining an accurate estimate of the k-processor run-time distribution. We show that the k-processor prediction method delivers accurate performance predictions and demonstrate the validity of our analysis on several robot motion planning problems. Key words: randomized path planning, randomized parallel search, performance evaluation, parallel computers. 1. Introduction Among the many skills autonomous robots require to support their activities is the ability to plan the paths they must take while conducting those activities. Motion planning is the process of computing paths that will allow a robot to move to different positions in its environment without hitting obstacles. Many motion planning algorithms have been developed (Latombe, 1991), but most are never used in practice. Motion planning is PSPACE-hard (Reif, 1979), which implies that, as the number of joints of the robot increases, the problem quickly becomes computationally unmanageable. Parallel search algorithms have been shown to be effective for solving combina- torially explosive problems (Ferreira and Pardalos, 1996), in particular randomized search problems. When the search space grows as a small exponent of the prob- lem size, parallel