抄録
Recently, rescue robots are expected to perform the work on behalf of people in hazardous locations where people cannot enter easily. For rescue robots running at the rough terrain, rollover fall is fatal. Accordingly, various studies to reduce the burden of falling avoidance operation is proposed. However, in these studies, implementation is hard and estimation accuracy is reduced in the complex terrain. In this study, we propose an easily implementable and highly accurate system to assess fall risk using a 3D depth sensor. In the proposed system, we mount the depth sensor on the robot to reconstruct CG models of the environment and the robot in a virtual space. Normalized energy stability margin are calculated from the estimated points. We verify accuracy of the time of fall down calculated by the proposed system comparing to the existing study, by using a dynamic simulator for mobile robots.
