Vision based automated formation for multi robot cooperation

Abstract

In a multi robot system, robots are required to cooperate with each other, and therefore should have the ability to make their own decision based on multiple input sensors not only from the robots, but also from nearby robots. The task of carrying oversized objects of different shapes poses a challenge in selecting an appropriate group formation. Hence, the main objective of this project is to establish an algorithm that enables multi robot system to carry large load by automatically selecting the required group formation to successfully execute the task. At first, a robot will need to identify the shape of the object (oversized bar, rectangular, square or circular shapes). Then, the robots will form a suitable formation to carry the object. There are two main problems in this project. First, the capability of the robot to identify the shape of the object because the object’s image will be a bit skew form the actual shape, due to the slanting angle of the camera used to detect the shapes of the objects. The second challenge is maintaining the formation of the robots, while carrying the load on top of the robots, to a specified destination. A multi robot system, built in-house is used in the experiments to investigate the performance of the algorithm proposed. Algorithms implemented in this project are leader-follower and behaviour based strategy. One of the robots will operate as the command giver or the leader to the other robots. The algorithm consists of communication strategies and autonomous decision making capability. The robot will be communicating with each other using Xbee wireless modules and extracting the behaviour of the other robots. Sensors placed around the body of the robots are utilized to detect their relative distance, and hence, used to maintain their formation, so as to prevent the load from falling down. All the decisions are made by the robots autonomously via the onboard controllers. The multi robot system is shown to be able to autonomously determine the shape of the different oversized objects, thus appropriately change into formations capable of transporting large objects to a specified destination point autonomously, with no outside intervention.