The advancement of autonomous vehicle technology necessitates robust systems for real-time traffic sign recognition and obstacle detection to ensure safe navigation. This paper proposes a comprehensive autonomous vehicle (AV) simulation framework utilizing the Carla simulator, advanced image processing, and state-of-the-art machine learning models to enhance autonomous navigation capabilities. The system initiates by loading a specified virtual map within the Carla simulator and deploying an ego vehicle equipped with a camera sensor to capture environmental images. These images undergo preprocessing techniques, to ensure suitability for analysis. Traffic sign recognition system uses a pre-trained YOLOv7 model, enabling the AV to identify and respond to various traffic signals amidst occlusion, adverse weather, and variations in placement and size. Concurrently, a pre-trained Faster R-CNN model detects obstacles such as vehicles and pedestrians along the vehicle’s path. A dedicated distance estimation module further analyzes these preprocessed images to calculate the relative distances of detected obstacles using depth estimation techniques, providing crucial spatial information. The decision-making algorithm resolves conflicts between recommended actions from traffic sign recognition and obstacle detection modules, prioritizing safety-critical commands to ensure operational safety and efficiency. The models are stored on the network edge, facilitating efficient deployment and real-time processing. This framework demonstrates an effective integration of simulation, image analysis, and decision-making, advancing the reliability and performance of autonomous navigation systems.
Traffic Sign and Obstacle Detection for Autonomous Vehicle Navigation using Edge Computing
2025-03-20
526581 byte
Conference paper
Electronic Resource
English
A systematic study of traffic sign recognition and obstacle detection in autonomous vehicles
Emerald Group Publishing | 2024
|DSCC2017-5187 Traffic Sign Recognition in Autonomous Vehicles Using Edge Detection
British Library Conference Proceedings | 2017
|Sign detection for autonomous navigation
SPIE | 2003
|Autonomous Navigation and Obstacle Avoidance
AIAA | 2009
|