Controlling Robotic Swarm Behavior Using Artificial Physics and Real Time Kinematics

LT David Armandt, USN

Recent commercial developments in small, low cost real time kinematic GPS sensors enable position realization for light, mobile platforms with centimeter-level accuracy. With a high degree of positional confidence, we explore the feasibility of close proximity operations of unmanned systems in an outdoor, GPS enabled environment. A computer-simulated hookian spring force is used to control a robotic “swarm” of agents, a technique called artificial physics. The computer model applies a proportional spring constant from position information and the resultant force is applied to each agent respectively. We validate the model by comparing it to analytic solutions, then further refine the model by comparing it to field testing data. With an accurate model of the system, user-defined tasks can be tested in simulations and the same algorithm can then control the behavior of the robotic swarm in an outdoor environment.