A Taxonomy and Survey on Sensor Blockage-Induced Degradation and its Impact on Advanced Driver Assistance Systems

Advanced Driver Assistance Systems (ADAS) rely extensively on sensors to perceive and interpret their environment, enabling safe and efficient navigation, object detection or other functionalities. However, sensor blockage due to environmental contaminants such as dirt, rain, or snow poses a significant challenge, impacting sensor performance and consequently ADAS reliability. This paper presents a comprehensive taxonomy and survey on sensor blockage in ADAS, focusing on cameras, radars, and lidar sensors. The survey begins with an taxonomy overview of sensor principles used in ADAS and their vulnerabilities to blockage. Cameras, based on geometric optics, are susceptible to opaque contaminants like mud and transparent ones like raindrops. Advanced studies in computer vision and deep learning techniques are discussed for detecting and mitigating these contaminants. Radar systems, operating on electromagnetic waves, face blockage challenges from snow and water, affecting their range and accuracy. Detection algorithms and impact analyses highlight methods to mitigate these effects, crucial for maintaining radar reliability in adverse weather conditions. Lidar sensors, using laser pulses for precise distance measurement, are vulnerable to dust and moisture accumulation. Detection strategies such as total internal reflection and deep neural networks are reviewed, emphasizing advancements in contamination identification and cleaning mechanisms. The paper concludes with insights into the overall impact of sensor blockage on ADAS, stressing the need for robust detection algorithms and standardized testing protocols.