Navigation of robots in obstacle filled environment has been a challenging task. Motor slip and vibrations remains the biggest contributor in navigating a robot accurately to avoid obstacles. Increasing the number of sensors will help to make accurate decisions; however, binding the information from multiple sensors remains difficult because the robot could be confused with the contradicting signals from multiple sensors. In this research, a modified Pattern Association Network Controller (PAN-C) called Pengkelasan Automatik peNamaan Corak Asas (PANCA) is implemented for navigation of a mobile medium size 2-wheels robot. This robot is equipped with Infra-Red sensors and an on-board CMOS camera to sense the environment. Several Artificial intelligence (AI) techniques are implemented to allow the robot to explore the environment using its own synapse through the Hebbian rule. When the robot detects some obstacles, the AI techniques will relate the existing sensory pattern to find the shortest travelling time and avoid the obstacles. Currently, PANCA is able to fuse two different sensor types successfully.
Project Leader: Prof. Dr Riza Sulaiman