Intelligent Control And Navigation Of An Outdoor AGV

Piloting AGVs in an unstructured, dynamic and complex outdoor environment has become an emerging area of research. In this paper we describe the development of a fuzzy na tor to perform the local navigation and obstacle avoidance tasks for an outdoor AGV which is realised from a golf cart. The hybrid architecture incorporates a high level 'modeland-plan' and a low-level 'sense-and-execute' layer. Each of the complex local navigational tasks that need to be carried out is analyzed in terms of primitive behaviours and expressed as an aggregation of such behaviours. The fuzzy behaviours are synthesized based on appropriately fused sensory data received from the complementary sensor devices, especially, proximity sensors and a laser scanner. A fuzzy simulator developed using Matlab is introduced which is ca able of monitoring the vehicle performance gh travelling profile, heading angles, change in steering commands, as well as the logic patterns of the proximity sensors. A method to associate these logic patterns to linguistic variables is discussed. Lastly, enhancements to the simulator are presented.