CPS: Medium: Collaborative Research: Collective Intelligence for Proactive Autonomous Driving (CI-PAD)

该项目的目标是开发通过车对车（V2V）和车对网络（V2X）共享的实时态势感知。该方法是将传感器的感知与对其情况的解释相结合，以实现更安全的决策，并考虑到车辆与基础设施之间通信的局限性。支持自动驾驶和自动驾驶车辆的高速公路系统将包括基础设施传感器和车载传感器，它们之间具有大规模连接，并在整个交通网络中分布智能。由此产生的集体智慧是，自动驾驶汽车作为移动传感器，与固定的基础设施传感器一起相互增强，以构建实时交通图像。这种实时图像用于开发主动驾驶行为，优化交通流量，最大限度地降低事故风险。更广泛的影响包括集中指导那些对需要研究生培训的职业感兴趣的本科生，扩大他们在计算机和工程领域的参与。研究人员组织了一个跨学科的项目，涉及信号处理和机器学习、控制和优化、通信和网络科学。主动驾驶集体智能框架包括以下几个模块：1)场景构建，包括信号处理和机器学习，通过多模态多视角传感器构建驾驶环境的表示；2)情景解读，包括逐级的驾驶环境动态分析；3)决策，包括优化和控制，以支持安全的主动驾驶和优化流程；4)故障安全网络，由通信和网络科学组成，支持在正常感知和通信条件下优化交通流，并在感知和通信受损条件下调节流量。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The aim of this project is to develop real-time situational awareness that is shared via vehicle-to-vehicle (V2V) and vehicle-to-network (V2X). The approach is to combine the perception of sensors with interpretation of their situation to enable safer decisions, and take into account the limitations of the communication between vehicles and infrastructure. A highway system that supports autonomous and self-driven vehicles will include infrastructure sensors and onboard vehicle sensors, with massive connectivity among them and distributed intelligence across the entire transportation network. The resulting collective intelligence is one where autonomous vehicles serve as mobile sensors that augment one another along with fixed infrastructure sensors, to construct a real-time picture of traffic. This real-time picture is used to develop proactive driving actions that optimize traffic flow and minimize accident risk. The broader impacts include focused mentoring of undergraduate students who are interested in careers that require graduate training, to broaden participation in the fields of computing and engineering.The researchers organize an interdisciplinary project in signal processing and machine learning, control and optimization, communication and network science. The collective intelligence framework for proactive driving includes the following modules: 1) Scene Construction, consisting of signal processing and machine learning for constructing a representation of the driving environment from multi-modal multi-view sensors; 2) Situational Interpretation, consisting of driving environment dynamic analysis at progressive levels; 3) Decision Making, consisting of optimization and control to support proactive driving for safety and optimized flow; and 4) A Failsafe Network, consisting of communication and network science that supports optimized traffic flow under nominal conditions of sensing and communication, and moderated flow under conditions of compromised sensing and communication.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.