Robust Fusion of LiDAR and Wide-Angle Camera Data for Autonomous Mobile\n Robots

Abstract

Autonomous robots that assist humans in day to day living tasks are becoming\nincreasingly popular. Autonomous mobile robots operate by sensing and\nperceiving their surrounding environment to make accurate driving decisions. A\ncombination of several different sensors such as LiDAR, radar, ultrasound\nsensors and cameras are utilized to sense the surrounding environment of\nautonomous vehicles. These heterogeneous sensors simultaneously capture various\nphysical attributes of the environment. Such multimodality and redundancy of\nsensing need to be positively utilized for reliable and consistent perception\nof the environment through sensor data fusion. However, these multimodal sensor\ndata streams are different from each other in many ways, such as temporal and\nspatial resolution, data format, and geometric alignment. For the subsequent\nperception algorithms to utilize the diversity offered by multimodal sensing,\nthe data streams need to be spatially, geometrically and temporally aligned\nwith each other. In this paper, we address the problem of fusing the outputs of\na Light Detection and Ranging (LiDAR) scanner and a wide-angle monocular image\nsensor for free space detection. The outputs of LiDAR scanner and the image\nsensor are of different spatial resolutions and need to be aligned with each\nother. A geometrical model is used to spatially align the two sensor outputs,\nfollowed by a Gaussian Process (GP) regression-based resolution matching\nalgorithm to interpolate the missing data with quantifiable uncertainty. The\nresults indicate that the proposed sensor data fusion framework significantly\naids the subsequent perception steps, as illustrated by the performance\nimprovement of a uncertainty aware free space detection algorithm\n