Electronic cleaning glass for object detection in autonomous driving

This study proposes an electronic cleaning glass based on liquid dielectrophoresis for reliable operation of object detection systems in autonomous driving. The glass is capable of simultaneously removing both conductive and non-conductive droplets, as well as selectively removing specific types of droplets. The glass comprises a glass substrate and an array of transparent interdigitated electrodes coated with a dielectric layer. The contact angle of droplets with various dielectric constants is quantified in dependence on the amplitude and frequency of the applied voltage. The minimum switching time of the voltage for each interdigitated electrode is measured using a digital I/O board. Based on the aforementioned results, experiment is performed to remove multiple droplets with various dielectric constants simultaneously, and the cleaning rate is quantified over time. Furthermore, it is experimentally verified that the glass is capable of cleaning only certain types of droplets. Finally, the effect of the glass's operation on object detection performance is evaluated by measuring mean average precision (mAP) in images captured by the camera equipped with the glass. It is observed that mAP decreased by the contaminant but returned to its original value upon operating the glass.