An error detection model for ultrasonic sensor evaluation on autonomous mobile systems

This paper presents an approach to safe navigation of autonomous mobile systems within partially known or unknown dynamic environments. In this context, faulty, maladjusted and otherwise influenced sensors must be recognized (error detection), and adequate measures for failure correction (error recovery) must be taken. As a general basis for monitoring the state of environmental sensors, a so called "error detection model" was created, which consists of sub-models for data from laser range finders and ultrasonic sensors. With the aid of the created models, different kinds of redundancy can be utilized and consistency and plausibility checks can be carried out. The error detection model serves for failure recognition based on environment modeling and on hypotheses for expected sensor readings.