Autonomous navigation with teams of aerial robots

Abstract

There are many examples in nature where large groups of individuals are able to maintain three-dimensional formations while navigating in complex environments. This paper addresses the development of a framework and robot controllers that enable a group of aerial robots to maintain a formation with partial state information while avoiding collisions. The central concept is to develop a low-dimensional abstraction of the large teams of robots, facilitate planning, command, and control in a low-dimensional space, and to realize commands or plans in the abstract space by synthesizing controllers for individual robots that respect the specified abstraction. The fundamental problem that is addressed in this paper relates to coordinated control of multiple UAVs in close proximity. We develop a representation for a team of robots based on the first and second statistical moments of the system and design kinematic, exponentially stabilizing controllers for point robots. The selection of representation permits a controller design that is invariant to the number of robots in the system, requires limited global state information, and reduces the complexity of the planning problem by generating an abstract planning and control space determined by the moment parameterization. We present experimental results with a team of quadrotors and discuss considerations such as aerodynamic interactions between robots.