Position rracking in rough-terrain

Abstract

As explained in the introduction, accurate pose tracking is essential for applications where no absolute positioning means is available for extended periods of time. Such conditions hold for underground mining vehicles and planetary exploration rovers, for example. That kind of environment is a challenge for position tracking because of the roughness of the encountered terrain. Indeed, the vibrations, the high-amplitude orientation changes and wheel slip have a negative impact on the quality of sensor measurements and thus on sensor fusion. The previous chapters presented the different actions that have been taken to increase measurement quality. Chapter 2 introduced a platform leading to smooth motion across obstacles. That way, both the vibrations and wheel slip are limited. Then, Chapter 3 and 4 presented two complementary approaches for increasing the accuracy of the odometry. The smoother motion and increased accuracy of the odometry enable sensor fusion in favorable conditions. The intent of this chapter is to develop a method for combining different sources of information to estimate the six degrees of freedom of a rough-terrain rover. The approach, based on an extended information filter, is presented in Sect. 5.3. Section 5.1 gives a survey of the sensors that can be used in rough terrain and section 5.2 presents the problem of having sensors distributed at different places on the rover. Finally, Sect. 5.4 presents the experimental results, validating the theory. © 2008 Springer-Verlag Berlin Heidelberg.