Multi-robot Collaboration To Explore The Unknown Environment

In the 1980s, with the development in robotics, distributed artificial intelligence and distributed systems, the research of robot system has been directed to distribution systematization, and intelligence. The coordination problems among multi-robot systems have been extensively studied recently. The multi-robot systems with high flexibility and adaptability have various potential applied domains such as aerospace and aviation, military, industry, services and so on. Robots should be able to carry out the mission autonomously, where human beings can not enter, for instance, the hazardous mine, nuclear relics and reconnaissance under water. In order to improve the performance of multi-robot teams for exploring the unknown environment and building up the environment map effectively, the cooperating mechanism of multi-robot needs to be designed rationally, with appropriate algorithm for system optimization. This is significant to the realistic application.This paper focuses on the coordinating mechanism of multi-robot team exploring the unknown environment. At first, the original frontier-based algorithm will be introduced with the pros and cons. And then the improved frontier algorithm will be presented, which takes the orientation of frontiers into account as well as the distance of frontiers. It develops the exploring strategy in the basic algorithm and improves the team's performance. An Integer Programming model for the team objectives is constructed. Ant colony systems algorithm is applied to this optimization problem, as an approach for the system optimization of multi-robot team. The main study work and academic contribution are:(1) Instead of the complex information computing method in the utility algorithm, a straightforward method is presented for coordinating the behaviors between robots. It is called distribution function. It can be used to avoid the collision and conflict among the robots, independent on the environment structure.(2) Improved frontier algorithm is presented for reducing the repeated coverage, a common problem for most existing frontier-based exploring methods. It takes into account both the orientation and distance of frontier, decreases the probability of repeated coverage.(3) Optimize the team objectives of multi-robot team. The existing methods tend to assign task individually for robots using the Greedy algorithms to bypass the NP hard problems, which leads to imbalance of workload assignment for robots. This paper constructs an Integer Programming model for the team objectives, and uses Ant Colony System algorithm for optimization.We conduct simulations in different environments, and compare our improved frontier algorithm with the basic method. Results demonstrate the efficiency and advantage of our method.