Rough computational methods on reducing cost of computation in Markov localization for mobile robots

Markov localization can be applied to estimate a robot's position under global uncertainty. However, for larger maps the computation of the probability density in the global environment and maintaining it in real time is very costly. Analysis of the Markov localization algorithm reveals that much of the computation can be done in advance. We use rough computational methods to process environmental feature data and apply an incremental strategy in the algorithm to reduce the cost of computation for the robot's localization in real time.