Traffic congestion is a growing problem and is estimated to increase by 50% in the next 15 years without significant action [2]. Vehicle platooning is an exciting concept has been demonstrated on highways showing improvements in fuel efficiency, travel time, safety, and environmental impact. However, the high cost of infrastructure changes and in-vehicle technologies required does not justify about 10-20% [7] efficiency demonstrated to-date. Minimizing the inter-vehicle distanceand increasing speed of vehicles safely has the potential to double or even triple the efficiencies on both highways and city roads. But, unfortunately, the existing in-vehicle, V2V or V2I sensing technologies are woefully inadequate for these purposes (too slow or bulky or poor accuracy or expensive). The goal of the proposed research is to develop computer vision-based system for high-speed and high-precision estimation of inter-vehicle distances. Acomprehensive study will be conducted to evaluatethe impact of gap distance on traffic throughputand energy efficiencies. Based on the findings, we will develop a smart, cost-effective sensing and processing system that can work reliably for very short gap distances in all lighting and weather conditions. We already have collaborations with Traffic 21 (UTC TSET), NSF funded CPS project partners in RI and ECE (Hebert, Bagnell, Mertz), collaborations with industry (funding from Intel, Ford and likely GM), and endorsements for collaborations with Pittsburgh city and local partners like GASP and CMU CREATE Labs to study environmental effects (air quality) of traffic and industrial sources. The funding from Metro 21 will help put together a larger and longer-term effort in solving the transportation, environmentaland energy problems in America.
Background and Significance Traffic congestion instantly became a problem when mass vehicle production and urbanization took hold in the early 1900s. Despite efforts to reduce congestion by constructing highways and urban expressways, and widening roads, congestion is still a significant source of headache in cities of all sizes [1]. Traffic has a profound negative impact on productivity, fuel expenses, regional economic health, quality of life, and the environment. In 2013, traffic cost the U.S. economy $124 billion, which is estimated to increase to $186 billion by 2030 [2]. Additionally, commuters lost 5.5 billion hours, wasted 2.9 billion gallons of fuel, and created 56 billion pounds of excess CO2. To put those numbers in perspective, commuters in a small city like Pittsburgh, PA lost $1 billion, 46 million hours, 21 million gallons of fuel, and created 431 million pounds of excess carbon dioxide [1]. An ITS approach to addressing these problems was proposed by General Motors at the 1939 New York World’s Fair [3]. GM envisioned an automated highway system where vehicles could automatically drive on the road and follow other vehicles at safe distances in vehicle platoons. Current prototype deployments of vehicle platoons utilize local sensors such as cameras, LIDAR, RADAR, and GPS, DSRC for vehicle-to-vehicle communications, and sensors/computers built into the infrastructure [4, 5, 6]. Such systems are able to achieve platoons traveling at 50 miles per hour with 6 to 10 meters between vehicles. With this gap, fuel waste can be reduced by 10 -20% [7] and fatalities can be reduced by 10% [8]. Problem Definition Greater fuel efficiency and safety are possible with tighter gaps between vehicles. Additionally, small gaps are necessary in metropolitan areas where, unlike highways, vehicles need to be densely packed between intersections and traffic lights. Achieving tighter gap spacing (e.g., 1 meter), however, is very challenging for control systems using typical local sensors because of their slow acquisition rates and low accuracy (Table 1). Some of these sensors, such as GPS and RADAR, are also unreliable in cities where tall buildings are present. Cameras, however, have proven robust in a variety of road environments for many computer-assisted transportation systems ranging from complex systems such as autonomous driving to simpler advanced driving systems such as lane detection. But their accuracies and precisions are unreliable in unconstrained environments. Wepropose a fast vision-based platform with smart sensing and lighting for vehicle trackingand inter-vehicle distance estimation that has the potential to overcome the issues in traditional vision-based sensing. Approach to the Problem Vision-Based Platform:The requirements of the vision-based platform are fast and accurate estimation of the speed and distance of a vehicle in front—information needed to automatically follow the vehicle. The method we propose relies on projecting a fast modulating pattern or a set of patterns such as bar codes on the rear of vehicles, which makes it easier to identify and locate vehicles in frontand are not visible to the naked eye. Based on deviations from the expected parameters of the pattern, such as size, shape, and sharpness, speed and distance can be calculated. The platform will consist of a camera, processing unit (e.g., FPGA), and projector. The camera captures images of the road environment. The processing unit controls the system and processes the images captured from the camera. The projector displays the patterns. We have experience developing similar systems [10, 11], but expect this platform to require much faster performance. Therefore, as illustrated in Fig. 1, these components will be tightly integrated with custom hardware to ensure high-speed, high data throughput between components. We will use an FPGA for system control,image processing, and distance/speed calculations to take advantage of hardware acceleration and hardware parallelism. The FPGA board will also interface with the camera and projector for high-speed data transfer. Communication with a high-speed projector will permit dynamic adjustment of the patterns projected if necessary, for example, if the vehicle is too contoured or textured. The platform will also have an interface for outputting navigation data for use by other vehicle sub-systems. The FPGA platform also plays theimportant role of latency reduction -order of microseconds to send data through any kind of programmableplatforms. The short latency between the camera/projectorshould help to simplify the computations.We have previously built a low cost high-speed binary projector [11] and also have experience building sensors that function in the presence of ambient illumination [13].Another possibility is to install a high-speed binary display (any DMD used in HUDs will suffice), in the back of the cars that transmit 1024x768x1000 bits per second and with our short latency, to visually communicate between cars over long lines at high speeds and throughput. Based on previous work in the area, we expect the camera will need to be greater than 1,000Hz. High frame rate cameras are costly, but we expect to only utilize a small portion of the camera’s sensor. Our previous work [12] with cameras that dynamically adjust exposure times based on motion willprove beneficial. It will allow an ordinary cheapcamera to be used for our purposes. Comprehensive Study:The exact platform speed needed is unknown due to a lack of studies on gap distance impact on traffic throughput. To study the impact of gap distance on traffic throughput, we will use traffic modeling software such as TRANSYT or TSIS. Specifically, we will use the study to determine the gap distance needed to provide maximal benefit for fuel savings, improved safety, commute times, etc. This optimal gap distance will then be used to determine the speed and accuracy requirements of the hardware design. CMU Project Team A team of experts from multiple research fields including imaging and optics, computer vision, and embedded systems will conduct the proposed research. The team consists of Srinivasa Narasimhan (CMU, Robotics Institute), James Hoe (CMU, Electrical and Computer Engineering), Robert Tamburo (CMU, Robotics Institute). Each member of the team has strong experiences in transportation related applications as well. Budget $75,000 is requested for the duration of this award to provide support for a PhD student and purchasing simulation software and hardware required to meet the goals of the proposed research. Future Growth The results of our study will show that vehicle platoons in metropolitan areas is worthwhile endeavor for reducing traffic congestion, and our platform can lay the groundwork for new research and development on technology for vehicle platoons in metropolitan areas. For example, how do traffic signals need to be controlled for tightly packed vehicle platoons? How would navigation data from our fast platform be integrated with information from slower sensors? Vision-based sensors are the critical component in many advance automotive systems found in vehicles today including lane departure warning, collision avoidance, automatic cruise control, headlight dimming, blind spot detection, adaptive illumination, etc. We will develop algorithms for vehicle tracking, but the FPGA can be programmed and the hardware can be modified to work for any of these applications. Integration of our platform for any of these applications would increase response time by a significant factor.
Jan-Apr: Gap distance study Jan-June: FPGA board development May-July: Camera Interface June-Aug: Project Interface June-Nov: Software/Algorithm Development Sept-Dec: Performance Assessment Dec: Report
Solving or even easing traffic congestion is relevant to many agencies –private and public. Internally, at CMU, we believe this is relevant to Metro 21, Traffic 21, and Scott Institute for Energy Innovation. We already have significant collaborations with University Transportation Center (TSET) and Pittsburgh city (Debra Lam, Chief Innovation Officer) who endorsed our just-funded (1.4 million USD) NSF Medium proposal (Narasimhan is PI, Hoe is Co-PI) on fundamentally re-designing computer vision based cyber-physical systems for in-vehicle, V2V and V2I technologies. The PI Narasimhan also has significant industry collaboration with Intel, Ford and now discussions are underway with GM on adopting the Smart Headlight technology in vehicles.We plan to leverage these industry partnerships to integrate our system on a vehicle with autonomous controls over speed and steering or adaptive cruise control.
Upon completion of the proposed research, we expect to have 1) a prototype of the vision-based platform for vehicle tracking, 2) a completed study on gap distance requirements for the platform and impact on traffic throughput, 3) performance assessments of the prototype, and 4) a technical report of the research. The major tasks to be completed are listed below with timeline shown in table above: -Perform simulations for study on gap distance -Develop FPGA board, camera to FPGA board interface, and projector to FPGA board interface -Develop system control software and image processing algorithms -Assess performance characteristics (latency) and accuracy of the platform (distance/speed estimates) -Write report describing the research methodology and results obtained
Name | Affiliation | Role | Position | |
---|---|---|---|---|
jhoe@cmu.edu | Hoe, James | ECE | Co-PI | Faculty - Research/Systems |
srinivas@cs.cmu.edu | Narasimhan, Srinivasa | Robotics Institute | PI | Faculty - Research/Systems |
rtamburo@cmu.edu | Tamburo, Robert | Robotics Institute | Co-PI | Faculty - Research/Systems |
mpvo@cmu.edu | Vo, Minh | Robotics Institute | Other | Student - PhD |
Type | Name | Uploaded |
---|---|---|
Final Report | 79_final_report_platooning.pdf | Sept. 28, 2018, 5:21 a.m. |
No match sources!
No partners!