International Partner Search

Innovation & Technology Offer

Innovative artificial vision system and method for detection of anomalous objects (pedestrians or animals) on motorways or highways

Country of Origin: Spain
Reference Number: TOES20160411001
Publication Date: 7 May 2018

Summary

Two spanish universities have developed and patented an artificial vision system based on the detection and subsequent notice of unusual element presences in roads and highways, such as pedestrians, animates and inert obstacles, which may endanger the lives of the vehicles occupants on these roads and even those pedestrian and animals. This could help traffic authorities and users to avoid and reduced potential traffic accidents. License or commercial agreement with technical cooperation sought.

Description

A research group from two different spanish universities, whose research field is the computer language and science, has developed a solution for the detection of anomalous objects on motorways or highways, through a static vision system.



Control traffic on public roads using automated surveillance systems is extending progressively, due to the decrease of hardware cost associated with such systems. There are already numerous commercial products that address the problem of detecting vehicles traveling at excessive speed and discover vehicles that have jumped a traffic light. Such products are used in many cases to the imposition of economic sanctions violator vehicles. However, there has been not so much attention devoted to the detection of situations that are uncommon, but may represent a significant danger to the lives of people circulating on the roads. Among these situations is the junction of highways or roads by persons or animals, which besides not been authorized, has a high probability of causing a serious accident.



The present invention is encompassed within the field of security systems in transport infrastructure and traffic. Specifically, it focuses on the detection and subsequent notice of the presence of unusual elements in roads and highways, such as pedestrians, animates and inert obstacles which may endanger the lives of the occupants of the vehicles on these roads and also those pedestrian and animals.



This solution particularly pretends to detect pedestrian and animals wandering within the driveway of highways and motorways, or its surroundings. Thank his utilization, the traffic authorities may be warned or even the drivers directly by cell phones or others communication systems, so that the security level on that kind of roads would be increased. The systems consist of static video cameras, located in different points along the highway or road. This allows that the scene overview remains the same, what means important advantages to detect objects, in contrast to the dynamic perspective of the cameras integrated in vehicles.



Specifically, the system includes a digital video camera providing images to a CPU which implements moving objects detectors. This objects detectors send to a tracing objects module the information of such objects in each photogram. Based on this information the tracing module identifies the trajectory and speed of those present objects, and facilitates further the information to an anomalies detector, which classifies the objects into normal and abnormal/ anomalous. In case an anomalous object is detected (pedestrian or animal) the information about the issue would be sent to a communication module allowing the traffic authorities or drivers to be warned.



The expected cooperation partner is a commercial or government organization able to implement this technology to address improvements in road traffic management and security. The universities are open to a license agreement to materialize the technology transfer or a commercial agreement with technical assistance to arrange the acquisition of the system and collaborate in its installation/ implementation and further develop some technical aspects to fulfil all the requirements.

Advantages and Innovations

Advantages:



-There are already some existing solutions potentially similar to this system but with different features and purposes, that involve a lack of information, less performance or incompatibility with the requirements of the main problem to be addressed. For example, there are systems for the detection of objects in railways, but they distingue only between situation with or without objects, incapable to determine if those objects are anomalous or not. That is inadequate for highways where normal and anomalous objects coexist. Other solutions, like tracing vehicles system or oncoming traffic detectors, don not work for other forms like animals or pedestrians or have a less performance, which in the described situation means to assume too much risk.



-The technical requirements and devices of the solution are standard, easy to implement and not expensive. 

   

Innovations:



-A stochastic approximation algorithm is in place. It is capable to estimate in real time the average and covariance matrix from the color observed in each pixel in the video. 



-A dynamic tracing model is in place to incorporate information about the moving objects recorded in the scene, deleting them when they come out from the scene and adding new ones coming in. This model estimates the registered objects’ position and speed, and stores the previous time instants, in which the objects were identified, with the aim to eliminate them from the model when they are leaving the scene.

Stage Of Development

Field tested/evaluated

Requested partner

-Type of partner: R&D institutions, commercial or government organizations. 



-Field of activity: Traffic management and road management, traffic security, Smart cities promoters...etc



-Task to be performed: 



Arrangement for the acquisition of the license 



Adquisition and coworking for the implementation, adaptation to the requirements needed or improvement of the system through technical cooperation agreement. (Commercial agreement with technical assistance)