Innovative Interaction Approach in IMM Filtering for Vehicle Motion Models With Unequal States Dimension

Abstract

Robust and adaptive vehicle state estimation and tracking algorithms have become a very important part within the autonomous driving field. The family of interacting multiple model (IMM) filters has shown to provide very effective and accurate state estimation in systems whose behavior patterns change significantly over time. The main reason for the improved performance of IMM filters compared to single model approaches is the mode mixing, which constantly aligns the different models. This paper proposes an innovative way for the mode mixing, when the state-vectors of the models are of different size. The proposed mixing approach consists of two parts: firstly mixing the common states and secondly weighting between original and mixed states based on the model probabilities. Results from artificial simulations and real world measurement setups are shown to demonstrate the validity of the approach. Compared to previously suggested solutions, the proposed approach is more general and the overall complexity of the mode mixing step is reduced, which is the main contribution of the presented paper.

Vision-aided intelligent and adaptive vehicle pose estimation during GNSS outages

“…Accordingly, the vehicle's motion patterns can be categorized into three types: 1 for stationary, 2 for low steering, and 3 for high steering. In this paper, the criterion for the classification is defined as [26,27]…”

Vision-Based Real-Time Motion Estimation of Ground Vehicles from Aerial Imagery

“…These mixing probabilities are computed based on the model probabilities µ (i) , which are updated using likelihoods L (i) k of each filter, prior model probabilities µ (i) and the state transition probabilities π ji . In the last step which is a fusion step, a weighted sum of the updated state estimates of all model filters gives the final state estimate where the more likely filters modify the estimates of the less likely filters [5,7]. A complete cycle of the IMM filter with standard (equal dimension) KFs is given in [4,7].…”

“…The filter can provide highly effective and accurate state estimation in systems whose behavior patterns vary significantly over time. The main reason for the improved performance of the IMM filter [3,4] compared to the single-model approach is the mode-mix, which always coordinates the different models [5]. A detailed overview of IMM methods in target tracking is provided in [6].…”

“…However, mixing the estimation results of unequal dimension models requires special attention, and there is no specific or deterministic way to solve this problem. But, various methodologies are proposed [8,9,10,5] to deal with mixing the unmatched state variables. An innovative interaction approach for UDIMM filters [8] used in vehicle motion estimation is proposed in [5] by reducing the complexity of the mode mixing step.…”

“…But, various methodologies are proposed [8,9,10,5] to deal with mixing the unmatched state variables. An innovative interaction approach for UDIMM filters [8] used in vehicle motion estimation is proposed in [5] by reducing the complexity of the mode mixing step. [10] presents novel approaches for unequal dimension track to track fusion(T 2T F ) using Covariance Intersection (CI).…”