Pedestrian crash avoidance mitigation
This article may require cleanup to meet Wikipedia's quality standards. The specific problem is: Ref tags should be after punctuation, not before. (WP:PAIC) (November 2018) (Learn how and when to remove this template message)
Pedestrian crash avoidance mitigation (PCAM) systems (USDOT Volpe Center) also known as pedestrian protection or detection systems use computer and artificial intelligence technology to recognize pedestrians and bicycles in an automobile's path to take action for safety. PCAM systems are often part of a pre-collision system available in several high end car manufacturers, such as Volvo and Mercedes and Lexusand lower end cars such as Ford and Nissan but are not widely used. Currently, as of 2018 using 2016 data, more than 6,000 pedestrians and 800 cyclists are killed every year in the USA in car crashes. Effective systems deployed widely could save up to 50% of these lives. More than 270,000 pedestrians are killed every year in the world. An excellent analysis of technology capabilities and limitations is provided in Death of Elaine Herzberg. Pedestrian safety has traditionally taken a secondary role to passenger safety.
Typically, PCAM systems are part of the technology in self-driving cars such as the Tesla Autopilot and use an integrated forward-facing camera and radar or lidar system designed to help mitigate or avoid a frontal crash. However, PCAM technologies do not require self-driving technologies, just cameras and radar. Sometimes, these can be enhanced with the addition of low-light detection for pedestrians and bicycles. The Nissan Leaf provides some PCAM capabilities in a commercial vehicle, but most manufacturers do not. For the 2019 model year, two Subaru models – the Outback wagon and related Legacy sedan – will join the redesigned Forester and Ascent SUVs to include the automaker's EyeSight crash-avoidance system as standard equipment. Subaru was one of the earliest adopters of a PCAM system (EyeSight) which is in phase III. Key players operating in the global PCAM market as of 2016 include Audi AG, Volvo Car Corporation, Continental AG, Robert Bosch GmbH, Toyota Motor Corporation, ZF Friedrichshafen AG, Autoliv Inc., Mobileye N.V., Subaru Corporation, Valeo, Honda Motor Co., Ltd., General Motors, Ford Motor Company, Daimler AG, Denso Corporation, Nissan Motor Corporation, BMW AG, and Magneti Marelli SpA. The PCAM market in North America is projected to reach US$1,997.3 million by 2025. In 2016, the U.S. Department of Transportation's National Highway Traffic Safety Administration officially announced that automakers in the U.S. have to include the autonomous emergency braking system as a standard feature for all cars and trucks by 2022: this is a key component of PCAM. A detailed explanation for manufacturers offering emergency braking as part of a pre-collision system and often PCAM is provided in .
Under certain conditions, if the PCAM systems determine that the possibility of a frontal crash with a pedestrian or bicyclist is high, it prompts the driver to take evasive action and brake by using an audio and visual alert. If the driver notices the hazard and brakes, the system may use some sort of brake assist to provide additional braking force. If the driver does not brake in a set time and the PCAM determine that the risk of collision with a pedestrian or bicycle is extremely high, the system may automatically apply the brakes, reducing speed to help mitigate the impact or avoid the collision entirely if possible. Usually, this is a setting the driver must make to initiate earlier, but it can be the default.
In order to recognize a pedestrian, the computational system uses AI pattern recognition technology that typically uses machine learning and deep convolutional neural networks based on millions of images. In a simplified description, images from the car's camera and radar are compared to the prototypes stored in the computer. If a match is made and confirmed, the other systems in the PCAM are invoked. PCAM technologies can be improved with additional information from connected vehicles. A thorough description of the processes for pedestrian detection in about 2010 is provided in . AI technologies have improved dramatically since then, as can be seen in an update in May 2016.
PCAM systems as part of ADAS
PCAM extend the pedestrian safety systems achieved through pedestrian safety through vehicle design with automated ADAS. Volvo had the first automated braking system focused on other cars, but including pedestrians in 2009. The Insurance Institute for Highway Safety (IIHS) has published the results of their tests for pre-collision automated ADAS and determined a 50% improvement with automated braking. They did not provide separate information for pedestrian safety. HLDI, a part of IIHS, provides some evaluations of most of the main pre-collision ADAS. They found that Subaru's Eyesight I PCAM cut insurance claims by 31% and its version II, by 40%.
- Yanagisawa, M., Swanson, E., Azeredo, P., & Najm, W. G. (2017, April). Estimation of potential safety benefits for pedestrian crash avoidance/mitigation systems. (Report No. DOT HS 812 400). Washington, DC: National Highway Traffic Safety Administration.
- Lienhard Pfeifer. Shearlet Features for Pedestrian Detection. July 2018. Journal of Mathematical Imaging and Vision.DOI: 10.1007/s10851-018-0834-9