作者: M. Soika
关键词:
摘要: This paper presents a failure detection framework for external sensors on autonomous mobile robots. Each sensor's measurements make statements about the robot's environment. Whereas correctly functioning sensors' are consistent, readings of faulty will lead to inconsistencies within model A probabilistic formulation both sensor and environment allows quantitative examination these inconsistencies. The obtained consistency measure describes how well statement, corresponding reading single matches with generated by all other sensors. By combining measures over several statements, robot is able deliberate condition this sensor. approach provides mechanism, which suitable any unknown flexible in terms types. It was verified experiments ROAMER, using sonar integrated into occupancy grid representation.