A Robust Certainty Grid Algorithm for Robotic Vision

In this paper we describe an algorithm for fault-tolerant sensor mapping for robotic vision. Basically, we use a certainty grid algorithm to map distance measurements into a two-dimensional grid. The well-known certainty grid algorithm can tolerate occasional transient sensor errors and crash failures, but will fail when a sensor provides permanently faulty measurements. Therefore we extended the certainty grid algorithm by a sensor validation method that detects abnormal sensor measurements and adjusts a confidence value for each sensor. This robust certainty grid approach works with at least three sensors with an overlapping sensing range and needs fewer sensor inputs and less memory than other approaches. Our method supports also reintegration of recovered sensors and sensor maintenance by providing a measurement for the operability of a sensor. We also present a case study with an autonomous mobile robot that features the robust certainty grid algorithm in a time-triggered architecture.