Particle Filter Approach to Utilization of Wireless Signal Strength for Mobile Robot Localization in Indoor Environments

Abstract

For many autonomous robotic applications, the capability to simultaneously create a map of the environment while localizing its position within it is of critical importance. This is typically achieved by fusing odometry information from the robotic vehicle with information from landmarks detected within the environment. Indoor environments often have existing wireless infrastructure, which can be used as landmarks by estimating the distance between the robot and the access point. The most practical way to attain this is by measuring the decay of signal strength. However, radio signal strength does not predictably attenuate indoors as it does in open environments due to signal interference, absorption, and reflection from objects within the environment, inflicting unexpected amplification or decay at the receiver known as multipath interference. This causes erroneous distance estimations due to the unexpected changes in signal strength attenuation. In this research, models of radio propagation as it relates to the received signal strength indicator (RSSI) are explored along with localization techniques which utilize these models. For development and testing of RSSI-based localization techniques, a simulation method has been described which utilizes a Markov chain to provide realistic multipath interference on simulated RSSI data. Using this simulation technique, methods for simultaneous localization and mapping (SLAM) are explored. Due to the difficulty associated with modeling RSSI attenuation and distance estimation, a particle filter based SLAM approach is proposed and demonstrated.