Computationally Efficient Path-Following using Adaptive Color Models

The ability to follow man-made paths and roads is an important capability for a number of robotic tasks. To operate in outdoor environments designed for humans, autonomous robots must identify footpaths, and drive along them. In this paper, we describe a computationally efficient approach to identifying and following footpaths, using only a single camera. Our technique takes the robot's kinematics into consideration when planning the best trajectory to follow the path that it is on. We show that our approach is highly robust to visual artifacts such as shadows, lighting changes, ground texture changes, and occlusions. I. INTRODUCTION The ability to follow man-made paths and roads is an important capability for a number of robotic tasks. Au- tonomous vehicles must successfully identify the road on which they drive. An outdoor tour-guide robot must identify the footpaths on which it leads its humans. Once the road or footpath is identified, the robot plans and executes a trajectory, attempting to keep on the detected road. Often, these robots have an extensive range of sensors, and make restrictive assumptions about the environment that they will be operating in. In this paper, we present a robust method for following footpaths using only a single camera, and taking into account the robot's kinematics. Our system is able to automatically deal with a variety of path materials and textures, changes in lighting conditions, shadows, and occlusions due to pedes- trians. We make no a priori assumptions about the path material, or the surrounding environment, relying on an adaptive modeling system to dynamically maintain models of the path in front of the robot. Once areas of path are detected, potential robot trajectories are overlaid on them and scored for suitability. The most suitable trajectory is then executed by the robot. We are particularly interested in following paths on the campus of Washington University in St. Louis. These paths are typical of a university campus in the United States, and include a variety of construction materials, including con- crete, brick, and red clay. The paths are generally bordered with grass, dirt, bushes, or walls. Numerous buildings and trees cast shadows across these paths, considerably altering the perceived color of the path material. The paths also tend to be quite narrow (approximately 1.5m) and winding, often with sharp turns. Additionally, they are often congested