Cooperative Control for Localization of Mobile Sensor Networks

Abstract

In this paper, we consider the problem of cooperatively control a formation of networked mobile robots/ vehicles to optimize the relative and absolute localization performance in 1D and 2D space. A framework for active perception is presented utilizing a graphical representation of sensory information obtained from the robot network. Performance measures are proposed that capture the estimate quality of team localization. We show that these measures directly depend on the sensing graph and shape of the formation. This dependence motivates implementation of a gradient based control scheme to adapt the formation geometry in order to optimize team localization performance. This approach is illustrated through application to a cooperative target localization problem involving a small robot team. Simulation results are presented using experimentally validated noise models. Summary. In this paper, we consider the problem of cooperatively control a formation of networked mobile robots/vehicles to optimize the relative and absolute localization performance in 1D and 2D space. A framework for active perception is presented utilizing a graphical representation of sensory information obtained from the robot network. Performance measures are proposed that capture the estimate quality of team localization. We show that these measures directly depend on the sensing graph and shape of the formation. This dependence motivates implementation of a gradient based control scheme to adapt the formation geometry in order to optimize team localization performance. This approach is illustrated through application to a cooperative target localization problem involving a small robot team. Simulation results are presented using experimentally validated noise models.

A Variable Communication Approach for Decentralized Receding Horizon Control of Multi-Vehicle Systems

“…This allows the local solutions of the decentralized RHC be closer to the global centralized solution which is unique for all vehicles. This leads to smaller mismatch between predicted and actual trajectories of neighbors and hence according to stability condition (12) this improves the stability of the decentralized RHC. Therefore, theoretically, one can say that the multi-hopping can improve the stability of the decentralized RHC.…”

“…This leads to smaller mismatch between predicted and actual trajectories of neighbors, and according to stability condition (12) this will improve the stability of the decentralized RHC. Therefore, generally speaking one can say that a larger communication radius can improve the stability of the decentralized RHC.…”

“…Stability condition (12) implies that if the prediction mismatch is bounded then the stability is ensured. Reviewing the stability conditions derived in [1] and [2] confirms this proposition.…”

“…Therefore, theoretically, one can say that the multi-hopping can improve the stability of the decentralized RHC. However, practically multi-hopping induces delays and noise that will adversely affect the control algorithm [12]. Also, the effect of the higher computation effort must be considered like the case of larger communication radius.…”

“…Beforehand, it is important to find out how the communication can be performed to enhance the stability. The stability condition (12) says that the integration of mismatch between the predicted and actual trajectories of the neighbors over the prediction horizon must be upper bounded. Hence, in order to enhance the stability of the system, i ε should be reduced as much as possible and the communication must be appropriately performed in this manner.…”