State Estimation for Systems with Implicit Outputs for the Integration of Vision and Inertial Sensors

This paper addressed the state estimation of a system with implicit output. We formulate the problem in the so-called deterministic H∞filtering setting by computing the value of the state that minimizes the induced L2-gain from disturbances to estimation error, while remaining compatible with the past observations. To avoid weihting the distant past as much as the present, a forgetting factor is also introduced. We show that, under appropriate observability assumptions, the optimal estimate converges globally asymptotically to the true value of the state in the absence of noise and disturbance. In the presence of noise, the esimate converges to a nieghborhood of the true value of the state. We apply these result to the estimation of position and attiude of an autonomous vehicle using measurements from an inertial measurement unit (IMU) and a monocular charged-coupled-device (CCD) camera attached to the vehicle.