Robotic Pose Estimation via an Adaptive Kalman Filter Using State-varying Noise

Swarm robotics offers the promise of enhanced performance and robustness relative to that of individual robots, with decreased cost or time-to-completion for certain tasks. Having many degrees of freedom, the swarm related control and estimation problems are quite challenging, specifically when the solutions involve a large amount of communication among the robots. Under certain sensing modalities, direct measurement of robot orientation may either be not possible or require excessive sensor processing. In this paper, a novel method is presented to vary process noise intensity as a function of an estimated state in order to arrive at the hidden state robot orientation. Experimental results are provided, demonstrating the efficacy of the method as well as the error reduction relative to fixed-noise estimation.