A compact dual-conical mirror catadioptric sensor for wide-FOV 3D vision measurement

Wide-field-of-view (FOV) 3D measurement is pivotal for applications such as industrial inspection and robotic navigation, where conventional systems are fraught with trade-offs among FOV, system volume, and measurement accuracy. To overcome these limitations, this paper presents a novel catadioptric 3D vision sensor, which ingeniously integrates a dual-conical mirror with a single camera. We introduce a systematic design methodology that combines geometric modeling with multi-objective parameter optimization to navigate the trade-offs among these competing factors. This results in a compact sensor (Φ120 mm × 193 mm) delivering an extensive circumferential FOV of 308.33^∘. For enhanced robustness, we establish a generalized measurement model encompassing both forward projection and reverse reconstruction under general assembly conditions, coupled with a robust image unwarping strategy for feature extraction. Experimental validation confirms high precision: angular errors range from −1.38^∘ to 1.13^∘ (average −0.06^∘), and 3D reconstruction errors show a mean error of 0.50 mm (range: 0.38–0.65 mm, relative error: 0.34–0.42%) across a 90–190 mm working distance. These results collectively validate the proposed sensor as a powerful and effective solution for high-precision, wide-FOV 3D measurement.