Passive infra-red pedestrian detection and avoidance system

Abstract

A passive infra-red pedestrian detection and avoidance system and method for augmenting the operation of a vehicle on a roadway, especially for identifying potential pedestrian / vehicular collision danger for the vehicle in operation and adjusting the position and operation of the vehicle accordingly, includes at least one passive infra-red sensor array mounted on the vehicle in operative communication with an image processor tied into the operational system of the vehicle. The system detects, using thermal imaging and processing, the presence of people that may be in or laterally crossing into the travel lane of the vehicle. The image processor analyzes the detection of a human thermal signature and determines if the detected human thermal signature is moving, in what direction and at what speed, to assess any potential threat to the pedestrian or biker, and further whether any responsive action needs to be triggered in the vehicle's operation to avoid a collision.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS[0001]This application claims the benefit under 35 U.S.C. § 119(e) of U.S. Provisional Application No. 62 / 670,209, filed May 11, 2018, which is incorporated herein by reference in its entirety.FIELD OF THE INVENTION[0002]The present invention generally relates to augmenting the operation of autonomous vehicles, as well as manually-driven vehicles, and more particularly relates to a system and method for identifying situations that pose potential dangers to pedestrians from vehicles in operation, specifically between the vehicle(s) and pedestrian traffic, and for adjusting the position and operation of the vehicle accordingly to prevent harmful interaction between the vehicle and pedestrians, as well as warning pedestrians and bikers of an impending collision.BACKGROUND OF THE INVENTION[0003]Many companies are developing autonomous vehicles for commercial and personal use on existing roadways for a variety of applications, including but not limi...

AI Technical Summary

Benefits of technology

[0011]The present invention is generally directed to a passive infra-red based pedestrian detection and avoidance system for augmenting the operation of autonomous or driven vehicles on a roadway. In accordance with embodiments of the present invention, the system provides a superior approach to assist a vehicle or driver in detecting and identifying potential interactions between a moving vehicle, such as a vehicle travelling on a roadway along a predicted pathway, and pedestrian or biker traffic.

[0015]In accordance with the present invention, said vehicle response can take the form of providing direct input to a vehicle's “Driver Assist” steering system and / or automatic activation of a vehicle's braking system. The system is designed to be an adjunct to other sensors and guidance systems on a vehicle, especially during times of challenging ambient conditions, and as such enhances an autonomous vehicle's autonomous guidance system.

Problems solved by technology

Until guidance systems on-board autonomous vehicles match or exceed the perception and analytical decision-making ability of human drivers, there will be numerous ongoing daily situations which will frustrate the ability of a full autonomous vehicle to properly and dynamically respond to, or react to, its surroundings.

Moreover, until autonomous vehicles can safely rely upon existing operational systems and sensors for safe and efficient operation and eliminate essentially all risks, the public will continue to be hesitant to put full faith in true autonomous operation of such vehicles.

Indeed, “real-world” autonomous vehicular tests have shown to have had numerous guidance failures, accidents, etc., caused by vehicles, pedestrians, or bikers laterally crossing into a travel lane that existing guidance systems and sensors have either failed to adequately detect, recognize and / or react to these situations in a timely fashion, or have had a high rate of “false” indications.

Even with a plethora of guidance systems and sensors in both autonomous and driven-vehicles, there have been failures to identify and avoid interaction with pedestrian and / or biker traffic for various reasons.

While various optically-based automotive and / or autonomous guidance systems and sensors (e.g., video, LIDAR, etc.) are capable of performing well under ideal visual conditions, their capabilities can diminish to unusable levels under adverse ambient conditions, such as during or shortly after rain, snowfall, fog, etc., or when it is dark outside and in low-lighted areas of the roadway, or even when the pedestrian blends into the surroundings based upon clothing, skin tone, weather, sun glare, etc.

Additionally, while the existing level of “on-board” sensors, cameras, devices, and interfaces can alter an autonomous vehicle's driving characteristics to a limited degree (e.g., by braking for unexpected obstacles and / or other vehicles, or steering a vehicle if it drifts out of its lane), there is currently an inherent extreme deficiency with specifically assisting autonomous vehicles in the identification of pedestrian traffic and furthermore determining when said pedestrian traffic is poised to negatively interact with a vehicle, and further assist such vehicles in automatically taking action to prevent such negative interactions.

Even under ideal ambient conditions, the CAN bus of the vehicle must coordinate a considerable number of sensors, and requires extensive analytical processing power to merely allow the vehicle to try to identify what expected and / or unusual or unexpected objects are during operation, with very little time to actually detect, recognize and react appropriately.

Further, even if the vehicle is able to identify an object or situation, there is still a challenge with having the vehicle then attempt to figure out what the correct procedural response should be.

As there are almost an infinite number of potential scenarios which require an appropriate response by the autonomous vehicle, this again proves to be exceedingly difficult, if not impossible, given the limitations of having a constrained amount of on-board processing power and database size in the vehicle, as well as little real time available for a response.

Images