Incorporating world information into the IMM algorithm via state-dependent value assignment

We propose two methods of incorporating world information as modifications to the Interacting Multiple Model (IMM) algorithm via state-dependent value assignment. The value of a state is a measure of its worth, so, for example, waypoints have high value and regions inside obstacles have small value. The two methods involve modifying the model probabilities in the update step and modifying the transition probability matrix in the mixing step based on the assigned values of target states. The state-dependent value assignment modifications to the IMM algorithm are simulated and compared with the standard IMM algorithm over a large number of game player-controlled trajectories for obstacle avoidance, as ground truth, and are shown experimentally to perform better than the standard IMM algorithm in both target's current state estimation and next state prediction. The proposed modifications can be used for improved trajectory estimation or prediction in real-life applications such as, e.g., Air Traffic Control, ground target tracking and robotics, where additional (world) information is available.