Speeding up the Convergence of Real-time Search Through Changing Value-update Rules
Yue Zeng · Computer and Modernization · 2003
The heuristics is used to solve the shortest path.In solving many paths,as an improved A~* algorithm,learning real-time A~* (LRTA~*) is a real-time search algorithm.Based on dynamic system information,LRTA~* can converge rapidly the optimal path.In this paper,a new method to speed up its convergence through changing value-update rules is proposed.The experiment shows that it often converges suboptimal solution faster than LRTA~*.It is a better method to solve the satisfactory solution between O-D in a big route network.