Experiments with a rule-based robot palletiser

The Oxford Autonomous Guided Vehicle Project is taking an industrial robot vehicle and seeing how it can be made ‘smarter’. The project is proceeding along two main avenues: improvements in sensors and in sensor data-fusion; and improvements in planning [1]. Examples of sensor systems with which we are experimenting include the vehicle’s basic laser triangulation system, stereo vision, sonar, infra-red, and laser ranging. The planning stages are based around the world model (a geometric database) [2,9]; one simplifying assumption (that is reasonable for this vehicle) is that the world model is an accurate enough representation for the planning systems not to be worried by the sensor data itself. Thus the world model serves as a bridge from the data integration system to the planning systems.