Telematics for fleets and insurance started with essential questions concerning:
- geolocation of vehicles
- fuel level monitoring
- the number of kilometers on the dashboard
These solutions are now fully operational.
We then saw applications while driving:
- eco-driving tips to save fuel
- navigation, to save travel time or km
Again, these elements are currently being deployed. Similarly, eco-driving assistance applications are being deployed to try to reduce the cost of energy-related operations. But the artificial intelligence that can be embedded in telematics systems (device, smartphone) can provide many other services. In particular, it is possible to significantly reduce the number of road accidents, thereby reducing operating costs.
COST OF ROAD ACCIDENTS
Road accidents cost fleets dearly:
- repair costs, if the fleet is self-insured, insurance costs otherwise
- cost of the immobilization period of the damaged vehicle
- cost of non-working time of injured personnel
- cost of late delivery
- brand image cost
It is clear that if a solution exists to significantly reduce the number of road accidents, this directly reduces operating costs.
This is what we are talking about in this article.
HOW ACCIDENT HAPPENS
Accident can be seen as a « rare » event that comes from unsafe acts. Those unsafe acts sometimes lead to emergency situations, or near misses : accident is avoided but it almost happened. And for some emergency situations, accident happens, it may be severe, and even fatality.
This way of « thinking » the accident has been developed in what is called the « theory of risk », proposed by the researcher Frank E. Bird in the 70’s.
This theory presents the so-called « triangle of risk » :
Idea of NEXYAD was to detect unsafe acts, and then tell driver when it is time to come down and come back to a cautious behavior.
This was made through 12 international collaborative research programs, during 20 years, and in practice, it meant « putting the traffic regulation rules » into an AI.
Of course, our AI also integrates deep learning and reinforcement learning technologies, but the main core is a cognitive AI applying Fuzzy logic and Possibility theory.
This AI is what Americans call a « XAI » (eXplainable Artificial Intelligence).
ARTIFICIAL INTELLIGENCE TO REDUCE THE NUMBER OF ROAD ACCIDENTS
NEXYAD has developed a new kind of AI for road safety. This AI aggregates all the data available on board, and can diagnose at any time the adequacy of driving behavior in the context, in terms of road safety.
This diagnosis is made by the AI SafetyNex which has integrated the rules of caution of the driving license of about twenty countries: as soon as a rule of caution is not respected (e.g. slow down slightly at an intersection with priority on the right , or left), the AI realizes this and can tell the driver what to do. This explains why driving school companies are very interested in our SafetyNex AI.
When driver is informed, he can modify his driving behavior in order to return to safe driving. This completely changes the way driving risk is measured.
Indeed, attempts to measure driving risk have already been deployed in telematics:
- detection of brutal driving (severe braking, cornering)
- geolocated recording of risk areas called black spots.
The first approach was imagined more than 10 years ago and has been deployed by insurers and fleet managers. Its effectiveness is very low, because brutal driving is not synonymous with risk. A fluid driver who would always drive at a constant speed, 10km/h below the limit, but who would skip all the stops, would be a very well rated driver by this score. However, he is a very bad driver. In addition, severe braking can only be detected when it occurs, so too late to alert the driver and prevent him from taking risks. It’s not predictive.
The second approach comes up against the mathematics of statistics. If on a stretch of road there has been no accident in the last 5 years (frequency observed in the past = 0), does this mean that the probability (value with predictive power of the future) is zero? In any case, few people will dare to affirm it and commit themselves. In summary, a null observed frequency does not mean that the probability is null (impossible accident). We understand that we can just say that the probability is very low.
Furthermore, if a very busy thoroughfare has had 1 serious accident in the last 5 years, and given the number of vehicles that pass on this thoroughfare per day, the observed frequency of serious accidents over the 5-year period is generally order of 10-8, or almost zero.
We therefore have a zero observed frequency which does not correspond to a zero probability, and a so-called black spot zone which gives an observed frequency almost zero. Without doing math, everyone will be able to understand that this does not have much predictive power.
NEXYAD has worked with traffic police services in 19 countries, and the diagnosis is the same everywhere: black spot areas change location every year.
This approach is therefore also a dead end.
This is why NEXYAD’s artificial intelligence solution is both very innovative and very operational:
- driver is warned before encountering an emergency situation because his driving behavior is not careful and this is detected. You can drive recklessly and not have an accident, but sooner or later you end up having one. NEXYAD’s AI detects each unsafe act, and informs the driver that he should modify his driving behavior ( e.g. slowing down very gently). Our client BRIGHTMILE testified on our website about the effectiveness of our SafetyNex driving assistance solution, which reduces the number of road accidents by more than 25%. This drop in accidents significantly lowers operating costs.
- At the end of each trip, caution and risk statistics allow fleet managers to keep track of the improvement in the behavior of their drivers. This makes it possible to negotiate with insurers to further reduce operating costs.
APPLICATIONS AND FIRST DEPLOYMENTS IN INDIA
The artificial intelligence software brick SafetyNex, from NEXYAD, is currently integrated by the company MONTBLEU Technologies, which targets road safety markets:
- driving schools, which can use this technology in a MONTBLEU smartphone App, and which makes it possible to measure improvements in safe driving
- insurers who wish to include UBI clauses in their contracts and who therefore need to know how prudent the driving behavior of drivers is
- fleets: cars, trucks, two-wheelers, which need to reduce their operating costs
The MONTBLEU application is called ROAD and will significantly improve road safety in India.
NEXYAD technology is also being integrated into new vehicles, for three types of applications:
- the safety score, like the one offered by TESLA in the USA. The safety score is made up of statistics of caution and lack of caution (called driving risk)
- the safety coach, which is the driver’s alert application to help him better anticipate road difficulties, reducing the number of accidents by at least 25%
- the preventive ACC, which consists, instead of informing the driver that it would be necessary to slow down slightly to remain cautious in view of the context, to tell the vehicle itself because it is robotic. We can thus go from the preventive ACC to the autonomous vehicle whose level of caution is the key to explaining its behavior.
Professional fleets wishing to reduce their operating costs now have a new AI tool, integrated on each continent by talented companies that take into account the real context of operations in each country.
Similarly, non-professional fleets, made up of employees of large companies who come to work by car or two-wheelers, can benefit from this tool and reduce the number of accidents, within the framework of corporate social responsibility.
In addition, insurance companies all want to integrate digital technology into their mode of operation, by modulating their prices according to the behavior of the driver. NEXYAD’s AI is a simple response that is easy to deploy on mobile phones.
Finally, driving schools have a new generation of tools to help students progress and monitor them over time, possibly with the help of insurers.
The technology has been validated by insurers, telematics companies, fleet managers and car manufacturers alike.
Published in Telematics Wire