A terrain slope estimation scheme using infrared camera for planetary exploration rovers

Abstract

Planetary exploration rovers face severe energy and safety restrictions, which have a strong connection with terrain slopes. During a steep slope traverse, a rover consumes more power and is exposed to higher risks of getting stuck or of overturning. It is essential for a rover to autonomously recognize and avoid steep slopes for efficient and safe operations. Existing techniques (e.g. stereo vision) do not completely address challenges in planetary exploration, such as low-textured terrain appearance and computational resource limitations. This paper presents a novel slope estimation method using a monocular infrared camera. The proposed method estimates slope normals based on surface temperatures on two different slopes. The surface energy model is employed to correlate thermal properties to geometrical properties of the terrain. The idea behind this approach is that the solar radiation, which is a major energy source for terrains, can differ by time, slope angles and directions. The difference in energy input generates the gap of surface temperatures between target and reference surfaces, which can be remotely detected with an infrared camera. The proposed method avoids the problem of terrain appearance as it only uses temperature measurements, and is also computationally efficient thanks to efficient preprocessing. The algorithm is validated through simulations and outdoor experiments. The results show the effectiveness of the proposed scheme to estimate terrain slope normals solely from temperature measurements.