A circular interacting multi-model filter applied to map matching

The multi-model interacting algorithm, based on the Kalman filter, is defined in the linear domain. In this paper, we propose a multi-model interacting filter for the circular domain. The proposed algorithm is defined in a Bayesian framework with a von Mises circular distribution. It is used to estimate the direction of a vehicle and to define its dynamic behavior. The different models are a right turn and a left turn. The proposed circular interacting multi-model filter is applied to Map-matching. The filter processes the sensor heading measurements. For this application we assess the proposed filter for the change detection and identification of the road on which the vehicle is traveling.