The push towards deploying autonomous-driving capabilities in vehicles is happening at breakneck speed. Semi-autonomous features are becoming increasingly common, and fully autonomous vehicles at mass-market scales are on the horizon. Cameras are cost-effective sensors, so computer-vision techniques have loomed large in implementing autonomous features. In a vehicle, these techniques must function "in real time." Unfortunately, this requirement lies at the heart of a significant disconnect: when computer-vision researchers refer to "real time," they usually mean "real fast"; in contrast, certifiable automotive systems must be "real time" in the sense of being able to predictably react to input information (such as a detected pedestrian) within specified deadlines so that adverse outcomes (such as striking a pedestrian) are provably precluded. The goal of this project is to eliminate this disconnect. It will do so through research on several fronts. First, a real-time computer-vision programming framework will be created by extending OpenVX, which is a recently ratified standard intended for developing computer-vision applications for embedded systems. Second, new computer-vision algorithms that exploit the features of this programming framework will be created, and methods will be developed to transform existing algorithms to make them "real time" in a predictability sense. Third, an experimental evaluation of "real-fast" vs. "real-time" computer vision will be conducted using driving simulators, sub-scale autonomous vehicles, and advanced testing infrastructure at General Motors. While industry is pushing hard in the area of autonomous driving, autonomous vehicles will never become a common mode of transportation unless methods for certifying real-time safety are produced. This project will focus on a key aspect of certification: validating the real-time correctness of computer-vision applications. The results that are produced will be made available to the world at large through open-source software. This software will include the new programming framework to be produced as well as tools for validating the real-time correctness of applications developed using this framework. In this project, a special emphasis will be placed on outreach to girls and women, as three female graduate students will be involved in the project. Such outreach will include: events involving the Graduate Women in Computer Science (GWiCS) group at the University of North Carolina (UNC), which hosts an annual research symposium targeted toward undergraduate women and other under-represented minorities; Tar Heel Hack, a hackathon for local middle and high school girls; the UNC Girls Who Code Club, which provides local girls in grades 6-12 with a community for learning about computer science; and the UNC Computer Science Department's annual Open House and Science Expo. These events will include hackathon projects as well as demos of a driving simulator and a sub-scale autonomous car.
The goal of this project is to explore vehicle platooning at scale in Smart and Connected Communities. The approach is the development of techniques and models that provide incentives for vehicles to join platoons and maintain their platoon memberships. Connected vehicle technology helps in forming vehicle platoons (virtual trains of vehicles traveling with small gaps between them) with benefits including improved energy efficiency, increased road capacity, and enhanced mobility. Vehicles in a platoon may benefit differently depending on where they are in the virtual train (for example, the lead saves less energy compared to a vehicle in the middle). As such, some vehicles may not be willing to join a platoon, or stay in one as better opportunities for platoon formation arise. This project explores how platoons with these competing goals will be formed and controlled, so as to understand how to motivate vehicles to participate in them. The Mcity testbed is part of the validation of the research work, and workshops involving local high-school and college students are planned. A stable platoon structure does not contain any coalition of vehicles who could increase their individual utilities by trading their platoon memberships. Given the dynamic nature of traffic streams, forming and maintaining stable platoon structures is a complex task, and requires accounting for both local and network-level conditions at the time of platoon formation. This proposal introduces a general framework that enhances optimal-control-based trajectory planning models by enabling them to also account for network level traffic conditions. This proposal further integrates stable platoon formation into the enhanced trajectory planning models, enabling them to incorporate both local and network-level information to form behaviorally compatible platoon structures that stay stable in a dynamic traffic stream.
Each year the nation spends over $400 billion to power, heat and cool its buildings. Moreover, buildings are a major source of environmental emissions. As a result, even a modest improvement in energy efficiency of the nation's building stock would result in substantial economic and environmental benefits. In this project, the focus is on improving energy efficiency in commercial buildings because this sector represents a substantial portion of the energy usage and costs within the overall building sector. Enhancing the energy efficiency of commercial buildings is a challenging problem, due to the fact that centralized building systems -- such as heating, ventilation and air conditioning (HVAC), or lighting -- must be synthesized and integrated with individual inhabitant behavior and energy consumption patterns. This project aims to design, analyze, and test a cyber-physical and human-in-the-loop enabled control system that can drive sustained energy savings in commercial buildings. It brings together expertise in computational building science, eco-feedback, network theory, data science, and control systems to integrate physical building information and inhabitants with cyber (building-human) interaction models to enable intelligent control of commercial building systems. Specifically, this project will: 1) design an integrated cyber-physical system (CPS), called Building Information, Inhabitant, Interaction, Intelligent Integrated Modeling (BI5M), aimed at reducing energy usage in buildings; 2) assess the complex inter-relationships between and across physical building and inhabitant models, cyber building-human interaction and intelligent control models related to energy conservation behavior; and 3) empirically test and validate modules and the overall BI5M system at test-bed buildings on Stanford's campus and Google's office park.
This research incorporates measurement (geospatial building data, energy use data), dynamics (inhabitant social networks), and control (enhanced user control of: plug-load devices, HVAC, lighting) into the BI5M system. The BI5M system is centered on a cyber Building Information Management (BIM) model of the building, and will encompass rigorous systems engineering that will explore relationships across the cyber-physical domains and develop new insights for how the scientific principles of cyber-physical systems can be used to influence the energy efficiency of commercial buildings through both occupant behavior and intelligent control. By integrating physical building information and inhabitants with cyber interaction modeling, the research aims to introduce an integrated human-in-the-loop control paradigm for commercial buildings. In addition to a testbed and validated CPS system for commercial buildings (BI5M), this project targets fundamental knowledge on: ontological components required to integrate dynamic data streams and control information into static building models; complex socio-spatial structures of inhabitants; insights into how building-human and human-human interactions impact inhabitant consumption behavior; and new control models that leverage input on the energy usage, spatial, social and behavior dynamics of inhabitants. The educational impacts of this project will extend to participants (students, faculty, Google employees in the test-bed buildings), as well as a broader student population through the integration of key insights from this work into courses/projects at all three collaborating universities (Stanford, Georgia Tech, and Columbia). The project team will also disseminate results to practitioners/policy-makers working in the building management space through an Outreach Workshop. Additionally, this project will broaden participation in computing fields through a diverse team and by partnering with the Girls Who Code nonprofit to integrate project data sets and tools into their activities.