IMM Filtering with Randomized Turn Rate to track Maneuvering Target
2024 IEEE Aerospace Conference(2024)
Abstract
A tracking system can track targets efficiently, if motion models assumed by the tracking system matches the motion models of the actual target. As targets may move depending on the scenario, multiple motion models may be needed to track such targets. Hence, multiple motion models are required as model set, for state estimation. During the state estimation process using the Interacting Multiple Model (IMM), a suitable model dominates and state estimation is done as a probabilistic weighted sum of the estimated state of all models. In real world, targets can move with different motion models and it is very difficult to cater to all possible motion models in the model set of the tracking system. This may lead to model mismatch, which in turn will increase estimation error during maneuver. To overcome this limitation, in existing literature turn rate is computed using estimated parameters of target states, like target velocity and radius of turn. Then based on computed turn rate, turn models parameters are updated. On the onset or at the termination of the maneuver the estimates may not be accurate and in such cases, the computed turn rate using estimates leads to increased inaccuracies on output estimates. To deal with such inaccuracies, in the proposed approach a set of Coordinate Turn (CT) models are updated with randomized turn rate concerning the highest probable turn rate, with an assumption that the turn rate is Gaussian distributed random variable with zero mean. The highest probable turn rate is selected based on the highest probable turn model among all turn models. The highest probable turn rate model is called the principal turn model. In the proposed approach CT models are updated with randomized turn rates around the principal CT model along with the fixed sets of motion models, so that there is a high possibility that in the next scan one of the turn models will be closer to the actual target motion model as in general target motion is not abrupt. This in turn will reduce estimation error during maneuver. Monte Carlo simulation carried out in this paper to bring out the utility of the proposed approach compared to the conventional IMM (CIMM), turn rate estimation and model updation based IMM and Novel Variable Structure IMM (NVSIMM) for tracking maneuvering target.
MoreTranslated text
Key words
Turn Rate,Maneuvering Target,Interacting Multiple Model,Interacting Multiple Model Filter,Set Of Models,Interaction Model,Probabilistic Model,Tracking System,Highest Probability,Motion Model,Target State,Target Velocity,Model Mismatch,Root Mean Square Error,True Positive,Conditional Probability,Transition Probabilities,Measurement Noise,Position Error,Likelihood Model,Constant Velocity Model,Left Turn,Probability Matrix,State-space Model,Process Noise,Previous Scan,Filter Model,Polar Coordinates,Run Number,Adaptive Rate
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined