Low Cost Sensors Ego Localization with IMM Approach for unusual Maneuvers

Low cost IMU–Odometer–GPS ego localization for unusual maneuvers

“…These simple and linear models are used in optimal linear Kalman filters: this means, instead of building the merging algorithm on a single but complex vehicle model, a modular approach is adopted. Simple vehicle models are set and implemented in parallel [19,18,20].…”

New likelihood updating for the IMM approach application to outdoor vehicles localization

“…• Estimate fusion: The filter output estimateX k|k and its covariance matrix P k|k are computed from a fusion of the weighted state estimates. More details on this filter implementation can be found in [2]- [6].…”

“…Moreover, a Constant Stop (CS) and a Constant Rear (CR) models are used to describe stops and backwards driving situations, respectively. Details on these models are found in [3], [8] and [6]. Moreover, [8] and [9] present various ways to setup the IMM filter, in terms of Markov transition matrix and the noise covariance matrices.…”

“…Traditional IMM gives interesting results for outdoor vehicle localization, particularly when provided data are synchronized and available at the same frequency [2], [6].…”

“…However, using the traditional IMM algorithm for road vehicle localisation as presented in [3] or [6] can reveal important drawbacks coming from various sources of error. Firstly, noisy sensors data used as inputs can cause the system drift.…”