Research on Unmanned Vehicle Perception Decision-Making Based on Improved YOLO

In this study, a solution based on the improved YOLO model and multivariate normal distribution model is proposed to address the deficiencies of environment perception and decision-making capabilities of driverless vehicles in complex environments. An environment perception and decision-making system adapted to complex conditions is constructed through artificial intelligence and big data technologies, which significantly improves the performance and recognition accuracy of the vehicle. In response to the call of the Strategy for the Innovative Development of Intelligent Connected Vehicle Industry, this study provides a practical foundation for the development of the technology. In terms of environment sensing, this study applies big data technology to deeply process VOC and COCO sensor datasets, and combines statistical optimization with YOLO and multivariate normal distribution to improve the system's stability and accuracy in detecting key traffic elements. In terms of decision-making system technology, the performance of the decision-making system was comprehensively evaluated through a multi-dimensional performance evaluation system, including real-time, behavioral accuracy, and other metrics, and optimized for path planning and behavioral decision-making through machine learning. In the future, the research will focus on sensor technology, data processing algorithms and the integration of neural networks and artificial intelligence, in order to continuously improve the environment perception and decision-making ability of driverless cars, and contribute to the development of intelligent transportation systems.