Solving the local minima problem for a mobile robot by classification of spatio-temporal sensory sequences

The local minima problem occurs when a robot navigating past obstacles towa rds a desired target with no priori knowledge of the environment gets trapp ed in a loop. This happens especially if the environment consists of concav e obstacles, mazes, and the like. To come out of the loop the robot must co mprehend its repeated traversal through the same environment, which involve s memorizing the environment already seen. This payer proposes a new real-t ime collision avoidance algorithm with the local minima problem solved by c lassifying the environment based on the spatio-temporal sensory sequences. A double layered classification scheme is adopted. A fuzzy rule base does t he spatial classification at the first level and at the second level Kohone n's self-organizing map and a fuzzy ART network is used for temporal classi fication. The robot has no Frier knowledge of the environment and fuzzy rul es govern its obstacle repulsing and target attracting behaviors. As the ro bot traverses the local environment is modeled and stored in the form of ne urons whose weights represent the spatio-temporal sequence of sensor readin gs. A repetition of a similar environment is mapped to the same neuron in t he network and this principle is exploited to identify a local minima situa tion. Suitable steps are taken to pull the robot out of the local minima. T he method has been tested on various complex environments with obstacle loo ps and mazes, and its efficacy has been established, (C) 2000 John Wiley & Sons, Inc.