A Rapid Recognition of Impassable Terrain for Mobile Robots with Low Cost Range Finder Based on Hypotheses Testing Theory

Abstract

The recognition of impassable terrain plays a vital role in the motionplanning of a mobile robot, which generally relies on expensive sensors such as stereovision cameras. This paper proposes a rapid impassable terrain recognition algorithmbased on hypotheses testing theory using low-cost range finders with different diffusionangles. In this algorithm, a slope estimation model using two range finders mountedon different heights is first established, where the influence of the diffusion angle ofthe range sensor is considered. To deal with inaccurate measuring from low costrange finders, the hypothesis testing theory is then applied to judge whether there isan impassible terrain approaching, where the historical slope estimation results aretreated as a sample set of the same slope, and the judgement of impassible terrainis then made based on the sampling set rather than the concurrent slope estimation.So the robot is only required to count the number of slope estimation that supportthe determination of a terrain as being impassible, and the judgement is confirmedonly when that number is larger than a precisely designed threshold value. Thenthe stable recognition for impassable terrain would be acquired while the risk ofwrong judgement is limited. The experiments’ results indicate that this algorithmcan provide a reliable recognition of impassable terrain using lower cost range finderswith different diffusion angles with minimal computation.