Learning Visual Measurement Model for Kalman Filter with Deep Convolutional Neural Networks

In practical applications, estimating the state for mobile robots or autonomous vehicles is complex, and often depends on multiple sensors. Kalman filter (KF) is one of the widely-used temporal filters applied in sensor fusion. Even though a general KF is simple to use, it requires to specify a priori for the system model and measurement model in state-space forms, which is still challenging. Aimi...