Abstract: In recent years, group target tracking (GTT) has received widespread attention. Due to limited radar resolution, many existing approaches treat a target group as a whole and focus on the estimation of the kinematic state and shape of the target group, without directly considering the individual properties of each target within the group. Most of these approaches assume that the group shape is time-invariant or slowly time-varying. In practice, target groups influenced by task intentions may exhibit behaviors such as expansion and contraction, and their shapes are time-varying. Accurate tracking of target groups with time-varying shapes is challenging using these traditional approaches. To address this challenge, this paper proposes a tracking approach for target groups with time-varying shapes. First, based on a Gaussian process model in which the target group shape is represented by radial distances from the centroid to several boundary points, this paper incorporates radial velocities into the dynamic model of the radial distances. Accordingly, a new shape dynamic model is proposed to characterize the time-varying evolution of the target group shape using radial velocities, such as expansion and contraction. For different shape dynamic models, the corresponding models adopt distinct sets of radial velocities. Based on these shape dynamic models, the target group tracking approach is developed using interacting multiple model (IMM) framework to rapidly identify true evolution modes and match suitable tracking models. Simulation results demonstrate the effectiveness of the proposed approach in terms of tracking accuracy and shape estimation performance.

Similar articles