车辆ADAS技术多源信息系统设计

Design of Multi-Source Information System for Vehicle ADAS Technology

智能化、网联化是车辆未来发展的目标。需要依靠大量复杂的环境信息及高性能处理器来满足智能网联车辆对海量数据的时效性要求,以便为车辆高级驾驶辅助系统(ADAS)功能及自动驾驶功能提供实时、准确、可靠的决策规划依据。设计了1种车辆ADAS技术多源信息系统。该系统基于TI多核异构的高性能系统级芯片(SOC) jacinto6来实现整体ADAS技术的数据预处理、算法集成及综合控制功能,包含底层车辆音视频桥接(AVB)域间系统的车身信息采集、激光雷达的千兆以太网三维点云数据处理及单目相机的卡尔曼目标跟踪预测。测试结果验证了多种辅助驾驶功能在此SOC上集成实现的可行性,解决了以往单芯处理器以及分散独立ADAS处理能力低、处理速度慢、体系结构单一及ADAS缺乏平台化、集成化开发的问题。

Intelligent and networked are the future development goals of vehicles.A large amount of complex environmental information and high-performance processors are needed to the time-sensitive requirements of intelligent networked vehicles for massive data,to provide real-time,accurate and reliable decision-making planning basis for the advanced driving assistance system(ADAS)functions and autonomous driving functions of vehicles.A multi-source information system for vehicle ADAS technology is designed.The system is based on TI’s multi-core heterogeneous high-performance system-on-chip(SOC)jacinto6 to realize the data pre-processing,algorithm integration and integrated control functions of the overall ADAS technology,including the body information acquisition of the underlying vehicle audio video bridger(AVB)inter-domain system,gigabit Ethernet 3-dimensional point cloud data processing of LiDAR and Kalman target tracking prediction of monocular camera.The test results verify the feasibility of integrating multiple assisted driving functions on this SOC,solving the problems of low processing capability,slow processing speed,single architecture and lack of platform and integrated development of the previous single-core processor and decentralized independent ADAS.