This paper describes Atlas, a hybrid metrical/topological approach to SLAM that achieves efficient mapping of large-scale environments. The representation is a graph of coordinate frames, with each vertex in the graph representing a local frame, and each edge representing the transformation between adjacent frames. In each frame, we build a map that captures the local environment and the current robot pose along with the uncertainties of each. Each map's uncertainties are modeled with respect to its own frame. Probabilities of entities with respect to arbitrary frames are generated by following a path formed by the edges between adjacent frames, computed via Dijkstra's shortest path algorithm. Loop closing is achieved via an efficient map matching algorithm. We demonstrate the technique running in real-time in a large indoor structured environment (2.2 km path length) with multiple nested loops using laser or ultrasonic ranging sensors.

This paper presents a concept for fault detection and calibration of external sensors on mobile robots, based on a grid representation of the environment. For every grid cell consistency measures are derived by evaluating the information stored in them. The measures are then used to deliberate about the condition of the involved sensors. Thereby a mobile robot is equipped with the capability of detecting faulty sensors during its operation. The approach is suitable for unknown environments and is flexible in terms of failure and sensor types. Furthermore, it allows sensor calibration in unknown environments without any reference objects. The approach was verified by several experiments with the autonomous mobile robot ROAMER using ultrasonic sensors.

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