Robust Ego-Motion Estimation with ToF Cameras

摘要

This paper presents an approach to estimate thernego-motion of a robot while moving. The employed sensor is arnTime-of-Flight (ToF) camera, the SR3000 from Mesa Imaging.rnToF cameras provide depth and reflectance data of the scene atrnhigh frame rates.rnThe proposed method utilizes the coherence of depth andrnreflectance data of ToF cameras by detecting image features onrnreflectance data and estimating the motion on depth data. Thernmotion estimate of the camera is fused with inertial measurementsrnto gain higher accuracy and robustness.rnThe result of the algorithm is benchmarked against referencernposes determined by matching accurate 2D range scans. Thernevaluation shows that fusing the pose estimate with the data fromrnthe IMU improves the accuracy and robustness of the motionrnestimate against distorted measurements from the sensor.