The main problem in the guidance of Autonomous Mobile Robots is obstacle avoidance in unknown environment. This paper presents a novel technique for path fixing for obstacle avoidance using a framework of cellular logic. The images of the unknown environment obtained from local perception devices are scanned by an array of neighbourhood windows and processed by the techniques of contouring, skeletonizing and centroid determination. High risk regions and low risk regions are considered for nearby and distant obstacles. A framework of cellular logic is used to determine the probability of collision. The various possibilities are classified into equally likely classes. The optimal path is the one that has the common probability of collisions. The mission direction is accordingly fixed instantaneously to meet the challenge of obstacle avoidance.