Image Signal Communication and Sensing for Traffic Key Representation Prediction.

Autonomous vehicles are a pivotal technology in today's intelligent transportation systems, improving traffic conditions by enhancing road safety, reducing congestion, conserving energy, decreasing emissions, and boosting travel efficiency and comfort. The key to achieving these objectives is to enable autonomous vehicles to perceive their surroundings in real time, with accuracy and stability, and to make appropriate decisions and controls on the basis of this perception. This perception relies on onboard sensors such as cameras, radars, lidars, and ultrasonic sensors. Radar sensors, one of the most frequently used perception devices in autonomous vehicles, are known for their strong anti-interference ability, effectiveness under occlusion, accurate range measurement, and wide coverage. However, radar sensors are limited by low resolution, high cost, large data volume, and difficulty in identifying object types and colors. Therefore, compensating for the inadequacies of radar sensing technology and enhancing the perceptual abilities of autonomous vehicles are ongoing challenges. In this paper, we introduce a novel image signal-based perception and transmission technology, aimed at overcoming the limitations of radar sensing and improving the perception efficiency and quality of autonomous vehicles. This technology employs image signals as the information medium, transmitting road information from the perception end to the processing end via wireless communication, thus enabling road condition perception. Compared with radar sensing technology, this image-based perception transmission technology is more cost-effective, as it only requires standard cameras and eliminates the need for expensive radar equipment. It captures a richer array of information, including object shape, size, position, direction, color, and texture, rather than just distance and velocity. The processing is more flexible, utilizing existing image processing and machine learning technologies for image signal compression, encoding, decoding, recognition, and analysis, thereby enhancing the accuracy and real-time capabilities of perception. [ABSTRACT FROM AUTHOR]