Stabilization of MAV and MAV-UGV swarms based on a visual relative localization
The aim of this talk is to discuss stabilization, control and motion planning techniques for steering swarms of unmanned Micro Aerial Vehicles (MAVs) and heterogeneous teams of Unmanned Ground Vehicles (UGVs) acting together with MAVs. The presented methods based on a visual relative localization of swarm particles are all designed for utilization of multi-robot teams in real-world dynamic environments without any need for global localization or motion capture systems. This approach aspire to be an enabling technique for deployment of swarms of micro aerial vehicles outside laboratories that are equipped with infrastructure for robots’ precise positioning.
Beside basic technical details on the employed image processing system, bio-inspired swarming approaches and model predictive control based formation driving techniques with vision feedback will be introduced in the presentation. The presented swarm behavior is inspired by Reynold's BOID model with its three simple rules: Separation, Alignment and Cohesion. The resulting methodology integrates the swarming with the visual relative localization and with a stabilization and control mechanism respecting high dynamics of unmanned quadrotors.
As the second example of the vision based stabilization of groups of robots, an approach for coordination and control of 3D heterogeneous MAV-UGV formations under a hawk-eye like relative localization will be presented. The core of the method lies in a visual top-view feedback from flying robots used for the stabilization of the entire formation in a leader-followers constellation. A Model Predictive Control (MPC) based methodology is employed to solve the trajectory planning and control of a virtual leader into a desired target region while keeping the following vehicles in a desired shape of the group. The crucial task in this endeavor is to ensure a direct visibility between aerial and ground vehicles, which is necessary for the formation stabilization under the hawk-eye like concept.
The swarming and formation driving behaviors as well as the possibility of the swarm stabilization with the visual relative localization in control feedback will be shown in numerous simulations and experiments in outdoor and indoor environment. The experiments are inspired by search and rescue applications, where the formation acts as a searching phalanx, by surveillance applications, where the swarm has to cover a given area, and by deployment of swarms in tasks of environment monitoring.
