Autonomous RGBD-based Industrial Staircase Localization from Tracked Robots

This paper presents an industrial staircase localization algorithm based on RGBD data from a tracked robot. This situation is really challenging as the camera is placed close to the ground. Moreover, RGBD can be really noisy on sparse staircases. Contrary to existing works, our evaluation relies on ground truth data provided by a motion capture system. Our experiments suggest that our algorithm can robustly locate industrial staircase. We also propose a new framework to evaluate stair localization performance from RGBD data. The overall performance allows to safety control a robot to rally the staircase.