Bernoulli Box-Particle Smoothing for Detection and Tracking under Interval Measurement

In this work, we derive the Bernoulli forward-backward smoother for interval measurement using finite set statistics, and box-particle implementation was given. A box particle is a random sample that occupies a controllable rectangular region of nonzero volume in the target state space, which has nature advantage when dealing with interval measurement. While smoothing reduces estimation error by delaying report time. Compared with Bernoulli box-particle filter, the Bernoulli forward-backward smoother improves performance according to the simulation result.