Abstract: We present a scalable approach to optimizing robot control policies for a target collective behavior in a spatially inhomogeneous robotic swarm. The approach can incorporate robot feedback ...