Multimodal Sensor Calibration

Accurate perception in autonomous vehicles demands precise spatial alignment across cameras, LiDARs, and radars. This project develops a powerful, generalized calibration framework that fuses information from diverse sensor modalities to enable robust environmental understanding.

Evaluating competing lens models is a core challenge in camera calibration. This project proposes a model-free, high-accuracy calibration method using projected fringe patterns to reconstruct per-pixel viewing rays — serving as reliable ground truth for benchmarking standard approaches.