Wheel-Terrain Contact Geometry Estimation and Interaction Analysis Using Aside-Wheel Camera Over Deformable Terrains

Wheeled mobile robots (WMRs) have been proven to be quite competitive and useful in outdoor missions. However, they may face serious sinkage or slippage on deformable terrains, and even get stuck or damaged, thereby causing mission failure. To mitigate these risks, it is essential to closely monitor the robot's states and its interactions with soft terrains. In this letter, we propose an integrated wheel-terrain interaction detecting algorithm which can measure sinkage, wheel-terrain contact angles, and slip ratio simultaneously online. This vision-based algorithm requires only one fisheye camera as its input. It has strong versatility and can be applied to rovers with different suspensions or morphology variations. We use a single-wheel testbed to verify our proposed algorithm through various experiments. The algorithm achieves a frequency of over 30 Hz, while maintaining a high level of accuracy. The relative error of sinkage and contact angle measurement is around 2%, and the relative error of slip ratio is less than 0.7%. We also propose a semi-empirical model to estimate the wheel sinkage-slip ratio relationship. With its help, we can avoid potential risks to improve the safety and success rate of the robot.