Greg Grudic, Jane Mulligan, Michael Otte, and Adam Bates. 6th International Conference on Field and Service Robotics, 2007.

Summary. To navigate efficiently in new terrain an autonomous vehicle should be able to observe and learn perceptual models for identifying traversable surfaces and obstacles, to allow steering and planning in the near and far field. As the robot passes through the environment however, the appearance of ground plane and obstacles may vary, for example in open fields versus tree cover or paved versus gravel or dirt tracks. In this paper we describe a working robot navigation system based primarily on colour imaging, which learns sets of models online as it moves through the environment choosing whether to apply current models, discard inappropriate models or acquire new ones. These models operate on complex natural images and because they are acquired and used as the robot navigates, learning and evaluation must be possible in real time.