A Multi-Robot System (MRS) is a system composed of multiple interacting robots pursuing common goals. The interest in MRS is growing more and more due to the fact that in some relevant context, multiple robots are superior to a single robot system in performing coordinated tasks.
First of all, a task can explicitly require more than one robot to be done or to be completed in time, i.e. the exploration of a large area. Increasing the number of robots provides a clear argument of flexibility; more robots can perform the same task or can substitutes each other in case of failure, thus enhancing the robustness.
There are two main ways of achieving coordinated multi-robot behaviors:
Considering the ROBOSWARM scenario, task allocation techniques can be applied for coordinating the swarm by trying to assign to each robot the most suitable task.
Coordinated navigation and exploration tasks may differ in the way they are realized, but they have the common feature to require swarm members to coordinate their movements in unknown environment. The robots must be spread in the environment in order to collect as much information as possible. Cooperation is also used to localize each other and to fuse information acquired from the environment.
Coordination algorithms in ROBOSWARM scenario are developed to be use with robots equipped with a small number of simple sensors and work without any a-priori knowledge of the environment. The algorithms use only local communication frameworks and, in some cases, the robots communicate simply modifying the environment (indirect communication).
Each mobile robot is able to safely move in a dynamic environment, avoiding obstacle using sonar sensors, performing wall following, escaping form dangerous position.
Composing this elementary behavior Robots can perform a complex task, like patrolling the whole environment to find an object, clean a target area, clean the whole environment in a coordinated way, reach the base station to communicate with central server or to recharge batteries, etc.
Figures 1 to 5 illustrate different phases of the coordinated exploration and path planning exercise based on ROBOSWARM infrastructure.
Figure 1: Initial configuration of the swarm and ROBOSWARM infrastructure in the building.
Figure 2: Navigation graph created as a result of exploration and later used for navigation and path planning.
Figure 3: Example of data stored on RFID tags; RFID tags implement the nodes of navigation graph.
Figure 4 (simulation screenshot) shows how robots are able to coordinate patrolling task by exploring different rooms and ensuring that each area will be cyclically visited (each colored trail represents a robot path).
Figure 4: Patrolling trajectories of the swam.
In the Figure 5 a mobile robot computes minimal path to the goal area, using the navigation graph. The navigation graph is generated by swarm as a result of preliminary exploration phase.
Figure 5: Optimal path planning on navigation graph.