Hybrid inference for sensor network localization using a mobile robot

摘要

In this paper, we consider a hybrid solution to the sensor net-work position inference problem, which combines a real-time filtering system with information from a more expensive, global inference procedure to improve accuracy and prevent divergence. Many online solutions for this problem make use of simplifying assumptions, such as Gaussian noise models and linear system behaviour and also adopt a filtering strategy which may not use available information optimally. These assumptions allow near real-time inference, while also limit-ing accuracy and introducing the potential for ill-conditioning and divergence. We consider augmenting a particular real-time estimation method, the extended Kalman filter (EKF), with a more complex, but more highly accurate, inference technique based on Markov Chain Monte Carlo (MCMC) methodology. Conventional MCMC techniques applied to this problem can entail significant and time consuming com-putation to achieve convergence. To address this, we propose an intelligent bootstrapping process and the use of parallel, communicative chains of different temperatures, commonly referred to as parallel tempering. The combined approach is shown to provide substantial improvement in a realistic sim-ulated mapping environment and when applied to a complex physical system involving a robotic platform moving in an of-fice environment instrumented with a camera sensor network.