Autonomous Outdoor Driving.

Much work has been done in the area of autonomous outdoor driving. Batavia and Singh [3] navigate their robot in locally smooth hilly terrain and use a yawing SICK laser range scanner in a fixed pitching angle towards the the ground. They use an object vs. freespace classification for driving. Similarily, Patel et al. [20] use also a yawable SICK scanner to classifiy drivable surfaces. Their work focusses on controlling the yawable scanner to acquire the necessary depth information while driving. They also use local gradients to classify drivable surfaces.

A good overview of the state of the art in military context for autonomous navigation in highly unstructured terrain is given in [16]. Two classes of algorithms are discussed: First, obstacle avoidance using ladar (for short range) or stereo camera (for long range). Second, terrain cover classification using a stereo camera for colour analysis or ladar for range texture analysis. The advantages and disadvantages of each kind of sensor for different applications is discussed.