RESUMEN

This study develops a vision-based technique for enhancing taillight recognition in autonomous vehicles, aimed at improving real-time decision making by analyzing the driving behaviors of vehicles ahead. The approach utilizes a convolutional 3D neural network (C3D) with feature simplification to classify taillight images into eight distinct states, adapting to various environmental conditions. The problem addressed is the variability in environmental conditions that affect the performance of vision-based systems. Our objective is to improve the accuracy and generalizability of taillight signal recognition under different conditions. The methodology involves using a C3D model to analyze video sequences, capturing both spatial and temporal features. Experimental results demonstrate a significant improvement in the model's accuracy (85.19%) and generalizability, enabling precise interpretation of preceding vehicle maneuvers. The proposed technique effectively enhances autonomous vehicle navigation and safety by ensuring reliable taillight state recognition, with potential for further improvements under nighttime and adverse weather conditions. Additionally, the system reduces latency in signal processing, ensuring faster and more reliable decision making directly on the edge devices installed within the vehicles.

RESUMEN

Taillight shape in a vehicle provides an essential lighting signal that enables the vehicle to be seen from the rear at night, thereby preventing rear-end crashes. This study aims to investigate the effects of taillight shape on vehicle conspicuity, and proposes ergonomic taillight shape solutions to vehicle designers and manufacturers. Two complementary experiments were conducted to examine three types of taillight shapes at three design levels. The first experiment was designed to investigate the detection speed of a driver and the fixation duration and fixation counts on leading vehicles with different taillight shapes, based on an eye-tracking methodology. The second experiment was designed to investigate the dynamic visual searching performance of a trailing driver for leading vehicles with different taillight shapes, based on a visual search task. The experimental results indicated that a long line-shaped taillight (striplight) was the optimal ergonomic solution for enhancing vehicle conspicuity. Vehicles with an enclosed contour-shaped taillight were more salient than those with an open contour-shaped taillight. Moreover, the experience and gender of the driver and the vehicle-observer distance were found to be closely related to vehicle conspicuity, and therefore, must be considered by vehicle designers when applying a specific taillight shape design. This study provides insights into the taillight shape design that not only aid vehicle designers or manufacturers in enhancing vehicle safety but also enable potential vehicle buyers to choose a safe lighting system.

Asunto(s)

RESUMEN

This paper presents an approach that can be used to measure height of driver's eyes and rear position lamps from a video, i.e., two important metrics used to set sight distance standards. This data plays an important role in the definition of geometric design of highways and streets. Our method automatically estimates the camera pose with respect to the road. It then requires selecting two points to obtain the height. New vehicles tend to be higher and larger. Consequently, this information shoud be updated. This approach has been applied on a large panel of vehicles. Our method was evaluated on vehicle height measurements. Our results suggest that our method achieves less than 1.8 cm (0.7 in) mean absolute error. Our experiments show an increase in the height of driver's eyes and taillights.

Asunto(s)