Chapter 14 - Becoming Unstuck in Time

Abstract

For a mobile robot to operate robustly in real-world applications, the programmer cannot afford to think in sequential time. A robot should not simply read data, react to it, and read more data. When the robot is running under some conditions it may appear to be doing just this, but it must be capable of gathering information from different systems and perspectives and then applying the implied corrections to the current moment. This chapter presents the hypothetical case of an outdoor robot running with lidar and global positioning system (GPS). In the example it is assumed that the robot is approaching a tunnel where the GPS will drop out and the lidar must be used for navigation. There will always be some difference between the GPS and the lidar position indications. During the time when both systems are reporting, it is critically important to find the true position and heading because of the narrower confines of the approaching tunnel. The error caused by motion cannot be ignored. There are several levels of compensation that can be made. To some extent, this will depend on whether the lidar processing is done in the same computer as the odometry or in a separate computer. If computing for sensors is done in the same computer as the odometry, the system is said to be tightly coupled. If the processing is done in a separate computer, the processing is said to be loosely coupled.