Sponsor: Sarnoff Corporation

• Project duration: 10/1/2004 - 9/30/2007
• Total project (est.): $10,391,125
• Requested ATP funds: $5,091,633

According to the National Highway Traffic Safety Administration, more than 41,000 individuals die and 3.4 million others are injured annually in road accidents, resulting in more than $150 billion in lost wages and incurred medical expenses. Surveys indicate a growing consumer demand for improved safety features. A joint venture led by Sarnoff Corporation together with Autoliv Incorporated (Ogden, Utah) and the Ford Motor Company (Dearborn, Mich.) plans to develop an advanced vision-radar threat detection (AVRT) system, which will autonomously detect hazards, predict crashes, and deploy safety features in milliseconds, mitigating or even averting collisions even at high speeds. The auto industry has long sought a practical collision prediction system for cars, but with only limited success. The key hurdle is achieving a near-zero false alarm rate (FAR). To be really effective such a system must be able to take decisive actions - such as pre-arming or applying brakes or even deploying airbags - automatically. But taking such actions when there is no real threat will cause accidents, so the system must have virtually no false alarms. Both radar- and vision-based systems have been tried, but the former have problems with measuring the direction of a possible threat and the latter have problems with measuring the distance. The Sarnoff team proposes to merge the two approaches, drawing on the strengths and negating the weaknesses of each. The proposed AVRT will combine radar detection pioneered by Autoliv with vision-based systems developed by Sarnoff. The ARVT system will have enormous computational requirements and must be able to respond within a couple hundred milliseconds, so a key part of the project is to develop unique efficient point-tracking and parallelized algorithms that will produce a near zero false alarm rate (FAR) and a specialized microprocessor 256 times faster than existing Pentium 4-based hardware. Plans call for the system to eventually transfer to a single low-cost computer chip, enabling broad-based consumer deployment. Prior to product release, the partners plan extensive testing that models real-world conditions such as variations in weather, lighting, host and target vehicle velocity, host and target vehicle trajectory, and driver evasive measures. Ford and Autoliv both sought internal funding but could not because the required technological advances are much greater than is typical with this type of project. For similar reasons, Sarnoff could not obtain venture capital funding. ATP funding will enable the partners to develop AVRT to a point where Ford and Autoliv can comfortably pursue commercial product introduction. Without ATP funding, the joint venture will not pursue the project, potentially delaying introduction of this type of technology by several years and allowing a less advanced, foreign-developed system to reach the market first and become the industry standard. When fully deployed, AVRT could save up to 16,000 lives and prevent 600,000 serious injuries annually. It will also increase U.S. automotive market share by providing a significant advantage over foreign competition, reduce accident-related expenses and insurance premiums, and create a new product industry.

Active Project Participants

• Ford Motor Company (Dearborn, Mich.)
• Autoliv ASP, Inc. (Auburn Hills, Mich.)

For project information:

Mr. Thomas V. Lento, (609) 734-3178
tlento@sarnoff.com

ATP Project Manager

Francis Barros, 301-975-2617
francis.barros@nist.gov

This is the fact sheet for this project as it was announced on September 28, 2004.
Click here for the latest version of this fact sheet.
Visit the Advanced Technology Program Home Page