Motion planning with uncertainty: the information space approach

Presents a general approach to motion planning with uncertainty based upon the Bellman principle of stochastic dynamic programming. The authors introduce the information space, whose elements represent accumulated information about a system. To each of these elements corresponds a certain knowledge of the system, that takes the form of a probability distribution. By applying stochastic dynamic programming (DP), the authors generate optimal or suboptimal motion strategies, i.e. motion commands corresponding to current knowledge, whose execution gives the system the greatest probability of reaching a goal configuration.