Fault Detection in Mobile Robots using Sensor Fusion

Fault detection and isolation in mobile robots has become a challenging task primarily due to uncertain and dynamic operating environments. The design of model-based fault detection methods would not be a practical real-time solution in view of the dynamic and uncertain nature of the problem. Also, conventional single-sensor approaches have limitations in practical applications. In this paper, a method of fault detection and isolation (FDI) based on a multi-level data fusion and response (behavioral) analysis technique is presented. The proposed FDI scheme mainly consists of pre-processing, sensor-fusion, a conflict monitoring unit, a confidence level computation unit, a high-level information fusion unit and a fault isolation unit. The developed FDI method is implemented in a simulated robot environment employing IR/ camera fusion for navigation and obstacle avoidance. The fusion-based FDI method is tested under faults in camera and IR sensor. With the developed approach, faults are detected in a timely manner and isolated accurately. Also, with the incorporation of sensor fusion, reliable and accurate sensor information is adaptively fused and fault tolerance is achieved under camera/ IR sensor faults.