Vision-based guidance design from sensor trajectory optimization

This paper considers a vehicle guidance design to achieve given missions using a single vision sensor. An extended Kalman filter is used for relative state estimation, and the estimates are fed back to the guidance system. This paper suggests that including sensor trajectory optimization into the design improves guidance performance. The guidance design is formulated as a minimization problem for a weighed sum of guidance performance cost, control cost and estimation performance cost. The weight for the estimation performance cost is determined by using the estimation error covariance. The optimization problem is solved numerically using dynamic programming. Simulation results for three vision-based control applications of target tracking, obstacle avoidance, and formation flight are illustrated. Also, a guidance design based on one-step-ahead optimization is suggested.