Mobile Robot Rough-terrain Control (rtc) for Planetary Exploration

Mobile robots are increasingly being developed for highrisk missions in rough terrain situations, such as planetary exploration. Here a rough-terrain control (RTC) methodology is presented that exploits the actuator redundancy found in multi-wheeled mobile robot systems to improve ground traction and reduce power consumption. The methodology “chooses” an optimization criterion based on the local terrain profile. A key element of the method is to be able to estimate the wheelground contact angles. A method using an extended Kalman filter is presented for estimating these angles using simple onboard sensors. Simulation results for a wheeled micro-rover traversing Mars-like terrain demonstrate the effectiveness of the algorithms. INTRODUCTION Mobile robots are increasingly being developed for highrisk missions in rough terrain environments. One successful example is the NASA/JPL Sojourner Martian rover (Golombek, 1998). Future planetary missions will require mobile robots to perform difficult tasks in more challenging terrain than encountered by Sojourner (Hayati et al., 1996; Schenker, et al. 1997). Other examples of rough terrain applications for robotic systems can be found in the forestry and mining industries, and in hazardous material handling applications, such as the Chernobyl disaster site clean-up (Cunningham et. al., 1998; Gonthier and Papadopolous,1998; Osborn, 1989). In rough terrain, it is critical for mobile robots to maintain good wheel traction. Wheel slip could cause the rover to lose control and become trapped. Substantial work has been done on traction control of passenger vehicles on flat roads (Kawabe et al. , 1997). This work is not applicable to low-speed, rough terrain rovers because in these vehicles wheel slip is caused primarily by kinematic incompatibility or loose soil conditions, rather than “breakaway” wheel acceleration. Traction control for low-speed mobile robots on flat terrain has been studied (Reister and Unseren, 1993). Later work has considered the important effects of terrain unevenness on traction control (Sreenivasan and Wilcox, 1996). This work assumes knowledge of terrain geometry and soil characteristics. However, in such applications as planetary exploration this information is usually unknown. A fuzzy-logic traction control algorithm for a rocker-bogie rover that did not assume knowledge of terrain geometry has been developed (Hacot, 1998). This approach is based on heuristic rules related to vehicle mechanics. Knowledge of terrain information is critical to the traction control problem. An key variable for traction algorithms is the contact angles between the vehicle wheels and the ground (Sreenivasan and Wilcox, 1994; Farritor et al., 1998). Measuring this angle physically is difficult. Researchers have proposed installing multi-axis force sensors at each wheel to measure the contact force direction, and inferring the groundcontact angle from the force data (Sreenivasan and Wilcox, 1994). However, wheel-hub mounted multi-axis force sensors would be costly and complex. Complexity reduces reliability and adds weight, two factors that carry severe penalties for planetary exploration applications. This paper presents a control methodology for vehicles with redundant drive wheels for improved traction or reduced power consumption. In highly uneven terrain, traction is optimized. In relatively flat terrain, power consumption is minimized. A method is presented for estimating wheelground contact angles of mobile robots using simple on-board sensors. The algorithm is based on rigid-body kinematic equations and uses sensors such as vehicle inclinometers and wheel tachometers. It does not require the use of force sensors. The method uses an extended Kalman filter to fuse noisy sensor signals. Simulation results are presented for a planar two-wheeled rover on uneven Mars-like soil. It is shown that the wheelground contact angle estimation method can accurately estimate contact angles in the presence of sensor noise and wheel slip. It is also shown that the rough-terrain control (RTC) method leads to increased traction and improved power consumption as compared to traditional individual-wheel velocity control.