Mobile Robot Fault Detection using Multiple Localization Modules

Most applications in service robotics require that the position of the robot is accurately known. Faults affecting the localization system can thus have serious effects on the overall performance. This includes internal hardware and software faults, but external disturbances and faults from the surrounding dynamical and complex environment are even more common in service robotics applications. This thesis makes two main contributions. The first one is a method for detecting faults affecting the localization system of a mobile robot. Most fault detection systems work with detailed models at sensor level, where sensor data is processed to decide if the system is in a faulty state or not. While this is often a powerful approach, it requires reliable models of the environment, sensor noise and the robot’s motion. The proposed approach is based on the observation that most of the modelling required for fault detection is shared with robot localization algorithms. The problems of localization and navigation have been extensively studied in the robotics community, and there exist many reliable methods and robust implementations of such systems. By combining the outputs from several high-level localization modules, and hence avoiding working with raw sensor data and detailed models, it is possible to detect faults affecting the robot. In this thesis, a low complexity model of such a combined system is proposed, and a detailed discussion of the corresponding design choices is given. An Extended Kalman filter is used to calculate the posterior probability distribution of the outputs of the localization modules. The alarm decision is made based on the Mahalanobis distance of the innovations and a CUSUM test. This approach is very flexible and does not need direct access to sensor data, nor modification of existing localization algorithms. The proposed method has been implemented and tested on an ActivMedia service robot. Odometry and a laser based scan matching method, described below, were used as position modules. The experimental results show that the approach works. The second contribution of this thesis is a method to increase the efficiency of pointto-point search in a scan matching algorithm. Scan matching is a method to estimate the relative displacement of a laser-scanning sensor (light radar) between data acquired at two positions. Scan matching is a good independent complement to other sensors like odometry and sonars. Here, scans are matched by maximization of a score function. This function is calculated from the distance between every point in the scan to be matched and the closes point in the reference scan. Straightforward search needs as many checks as the square of the number of points in the scan. A method to reduce the search space is presented that significantly reduces the effort for score calculation.