A new analytical algorithm is developed to address the problem of joint sensor registration and multi-target tracking with varying target number and observation uncertainty based on the labelled random finite set (RFS...
详细信息
A new analytical algorithm is developed to address the problem of joint sensor registration and multi-target tracking with varying target number and observation uncertainty based on the labelled random finite set (RFS) and expectation maximisation (EM). A new complete data log-likelihood function is derived with the measurement and state RFS variables, and the EM approach is employed to jointly estimate the sensor biases and target states. Moreover, a recursive implementation is provided to deal with the measurement accumulation, and the situation of the time-varying biases is handled. Since the estimates of the bias and state are analytically calculated, the performance of the proposed method is better than that of the traditional methods. The effectiveness and superiority of the proposed algorithm are verified using numerical simulations.
暂无评论