Focus

Data Analytics for safety of platooning vehicles

The development of automotive cooperative systems has progressed significantly to the point of real world testing (e.g. Google Driverless Car), and are anticipated to offer significant safety benefits for society in general. Most of these automotive cooperative systems are based in V2V and V2I to facilitate communications among vehicles and infrastructure, safety applications to analyse data and identify potential collisions, intelligent Human-Machine Interfaces (HMI) that warn drivers and true by-wire systems that can take control of the vehicle to avoid accidents and incidents.
The following benefits are expected. Reduced driver stress. Reduce the stress of driving and allow motorists to rest and work while traveling. Reduced driver costs. Reduce costs of paid drivers for taxis and commercial transport. Mobility for non-drivers. Provide independent mobility for non-drivers, including disabled people, and therefore reduce the need for motorists to chauffeur non-drivers, and to subsidize public tranport. Increased road safety. Reduce many common accident risks and therefore crash costs and insurance premiums. Reduce high-risk driving, such as when impaired e.g. by alcohol consumption. Increased road capacity, reduced costs. By allowing platooning (vehicle groups traveling close together), narrower lanes, and reduced intersection stops, reducing congestion and roadway costs. Increased fuel efficiency and reduced pollution.
The goal of the research is to demonstrate how we can apply and extend safety assurance frameworks to automotive cooperative V2V-based systems with platooning.
Consider a scenario where a long platoon of vehicles is travelling along a motorway. Suddenly the brakes fail on one of the car, so a Control Loss Warning (CLW) should be sent to following vehicles. The vehicle involved broadcasts this CLW to surrounding vehicles and infrastructure through Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communications (based upon 802.11p protocol). Upon receiving notification of the control loss event, the receiving vehicle and infrastructure determines the relevance of the event and establishes appropriate actions.
Vehicle platooning is a typical example of how safety and optimization of cyberphysical systems under performance/safety constraints constitutes a formidable problem. When performance and safety issues are concerned, the largest part of the current approaches relies on bruteforce analysis via simulation and/or emulation. In order to allow performance modelling in more general conditions, this use case will investigate machine learning tools to discover the safety performance trade-off on the basis of monitored data. The idea is to drive safety and performance analysis in an automated way, possibly via intelligible rules while preserving the model accuracy.
The CLW application will then be implemented on commercial off-the-shelf mobile robots to provide a readily understandable physical instantiation of the results of system failures with and without CLW on vehicle platooning in a lab context.


info: maurizio.mongelli@ieiit.cnr.it