Onboard telematics now can measure behavior of a driver, and therefore, car insurers

have early started this adventure of connected car, more or less successfully.


The simplest applications that have been deployed are:

. locate stolen vehicles

. measure the usage of the driver, and in particular the number of kilometers traveled in order to propose adaptive pricing (vehicle

  that always stays in a garage will never have an accident!)


But the main business of the insurer deals with the concept of risk, and then, we have seen a lot of telematics firms proposing automatic detection

of risky behaviors.

The most common is the so-called detection of "severe braking," which is based on the assumption that severe braking

reveals a lack of anticipation, and thereby a dangerous driving.

We now know that this assumption is totally false [1], but it is still in the mind of some insurers that « want to believe » there is a simple way to
classify human beings behaviours.

However, the lack of results of these deployments has led some German and US insurers to abandon Telematics [2].


The company NEXYAD has demonstrated that it is possible to « measure » in real time the risk of driving, and this stimulus now  keen interest in telematics
among insurers worldwide. This interest was even increased as NEXYAD won the BMW Tech date challenge with their onboard risk assessment App SafetyNex [3]

SafetyNex works where all other systems fail, simply because the problem was treated in a completely new way, without any « science fashion » consideration,
especially about the deep learning (or machine learning).


Indeed, the difficulties of developing  an application of efficient onboard risk estimation are :


. Science and facts : an accident is a rare event and inexplicable (definition :  "happens by chance", a driver has got one accident
  every 70 000 km, on average, most of which are harmless). Observing a driver during 5 years (to make sure an

  accident occured ) is long and ineffective (one accident does little to make individual statistics), and  variability factors of "road life situations" are extremely
  numerous, so it would take millions of drivers during decades before having relevant statistics.


. Ethics : driving behavior in itself has absolutely no direct link with the risk [1] (indeed we conceive  easily that drifting demos on an abandoned airport or just in front
  of a school at noon, corresponds to very different risks although the driving behavior is the same : it obviously needs to be "contextualize"). Contextualization (that is
  not present in the  "severe braking" experiments mentioned before) therefore demand to know, among other things, speed of  the vehicle, and where this speed is practiced.

  But as digital maps have all recoded the maximum authorized speed, then if you record the speed and geolocation in a cloud ... it potentially saves violations of speed limits.
  In many countries, ilcuding France, it is prohibited to record infringement to the law by non accredited organizations (like Insurance Companies). This totally disqualifies

  telematics boxes that record raw data in the cloud !

  However, some European insurers continue to test this kind of solution in the hope (bu in vain) that the "deep learning" and "data scientists" give their risk scores.

  But in any case in France (40 million vehicles market), the violation of the Penal Code is sanctioned and generally pursued by the CNIL [3 bis]. And insurance companies won’t
  have the opportonity to defend themselves saying ‘there is no choice » because SafetyNex estimates risk of driving without recording ANY confidential data !
  And it was shown that SafetyNex delivers every needed data to insurance companies (without any violation of driver’s privacy).


We can see with these two constraints that the solution of " big data statistics in the cloud using machine learning" can not be applied:

. statistics (or deep learning etc): accident is rare soi t won’t work at the individual level

. in the cloud: this is contrary to the laws that protect privacy of people.





The accident is a rare event (1 accident every 70 000 km on average and mostly minor).


This means that one must observe tens of thousands of km to observe ONE accident ... so to observe million accidents

(For statistics, you need millions of data), one must observe a huge number of kilometers traveled ... And in every place (because one area may be dangerous because of the presence of ravines,
another one because many roads intersect, etc’s never the same).


And as the observation of an accident is not enough, we must record all the measurable "variables" or « factors » (speed, acceleration, etc.) that describe

the behavior of the vehicle at the time of the accident (in order to "explain" the accident as say the statisticians).

Of course, nobody has millions of observations of accidents at each location of the infrastructure, then statisticians will just tell you « not enough data ».


What is the impact of unsuffissant volume of data on deep learning [4] ?  Well let’s take an example:

Let’s record for one person during 5 years (the time it taked to get an accident), the day of the week, the time slot, and driving signals (speed, acceleration, braking, ...). The result of this observation

of 5 years (it's long ? isn’t it ? ) will lead on average to 99 999 km without acident and 1 km where an accident occured.
Let’s say it was a Thursday, at 15:00, the vehicle was traveling at 100 km / h, etc.

As the vehicle drove frequently slower and faster than 100 km / h, the influence of the speed in the deep learning will be close to zero. However, the driver has never had an accident on Monday, Tuesday, Wednesday,
Friday, Saturday, Sunday.

Certainly, it has led many Thursday without accidents, but the only day he had an accident was on Thursday: the probability of having an accident on Thursday is therefore greater than that of having an accident the other days.
Here is what data analysis, statistics, or deep learning will conclude ù


Everyone can understand that this conclusion is completely wrong, and that it you observe that same driver for thirty years (the time to get several accidents), then you will see that the day is not a key factor (it can have
an influence if traffic varies with day, but obviously it os possible to have an accident any day of the week!).


As a conclusion let’s keep in mind that It's easy to make global statistics of accident on a large population (France, Europe, USA, ... hundreds of million people). But at the individual level, it is not so easy.

But the goal of onboard telematics is precisely to estimate a risk, at the local and individual level !


We know it seems obvisous when we say that it is not possible to study rare events without prior knowledge using data oriented mathematical methods (because there are few data and because those methods refer to the
« law of large numbers »). But it's better to say it because it is apparently not obvious to everybody (and it sounds always « cool » to tell your boss and your friends that you work on a deep learning application ! ^^).


SafetyNex circumvented this problem by working in a much more rational and finally "classical" way, using knowledge and risk evaluation methods already validated by experts of accident.


Note: To develop the theory of relativity, Albert Einstein did not record hundreds of billions of data to feed a deep learning system that automatically found the law E = mc2.
He used the knowledge of physicians, and inference methods of mathematics that have led to this formula. And then, in order to validate this formula, experimental physicists have performed hundreds of experiments.

It is exactly this approach that has been applied to develop SafetyNex: there are dozens of experts working on road infrastructure "risk diagnoses." NEXYAD worked in contact with these experts for 15 years (through collaborative
research programs PREDIT [5]) and developed SafetyNex which is a "knowledge-based system" [6], validated a posteriori on about 50 million km.

These experts work the same way than industrial risk experts in factories with methods like FMEA [7].


The difficulty of developing a tool like SafetyNex lies not the "technology" (gradual knowledge based system and possibility theory) because hundreds of startups in Silicon Valley (for instance) perfectly know these
techniques, but it resides in the extraction of deep knowledge of dozens of experts (that not always agree with each other, etc.). This extraction was made possible thanks to the collaborative French National Research PREDIT
projects « Arcos » and  « Sari ».

This research showed a key concept in accident: the "near accident" or "quasi accident" [8], a more regular notion than accident (And therefore a notion that can be studied mathematically).

Basically, if you put your feet in the water and strip the electric wires of the light of your ceilling, you are 100% in state of near-accident.
Note that you can do so without being electrocuted. It is the repetition of the act that eventually, randomly, will cause electric shock.


This concept is particularly interesting for the insurer because it measures the RISK THAT THE INDIVIDUAL TAKES, out of luck or bad luck consideration. It is exactly what the insurer needs to know.

And it is completely knowledge based : IF you put your feet in the water AND … THEN you are 100% in near accident case.

You do not need deep learning, you "know": SafetyNex works like this.


The advance of NEXYAD on this subject is so huge because extracting knowledge of dozens of experts in road safety in Europe, gathering experts when they disagree, etc ... is an incompressible duration,
whatever the financial strength of the company who wishes to do it.
SafetyNex applies about 5000 cause effects rules, and is usable WITHOUT DELAY: no observation period or learning period, when the driver dstarts driving with SafetyNex you know the  risk he/she takes iatevery moment.


Among these high-level expert knowledge is included the fact that 75% of accidents are due to inappropriate speed of the car to the danger of infrastructure.

Everything other factors (poor visibility, not compliant interdistances, rain, etc ...) are important, of course, but they explain 25% of the variance of the phenomenon.

When comparing SafetyNex to the work of the entire automotive industry (driver assistance systems with obstacles detection, etc ...) we can see that NEXYAD is the only partner who offers a tool that copes with main
factor of accident . All others are within remaining 25%.





Risk estimation in driving requires having contextualized synchronized data : how the driver drives, and where it takes place.


Now, as we have explained above, the recording of those data is in contravention with driver’s privacy because when you know how fast a vehicle drove
and where, you just have to read the speed limit on a regular electronic map, and then you know every infrinngement. Is it the job of the police, not of
insurance companies.


And note that, on the one hand it is forbidden to record such data in many countries, but on the other hand, it is totally unnecessary to [9] insurer. So it
shouldn’t happen !


SafetyNex bypasses this difficulty by performing all risk computing locally on the smartphone microprocessor, so that no indiscreet  data is recorded on the cloud.

Raw data are indiscreet : they may let easily know if you cross speed limits, but ot also let know who you you visited, when, etc … They are needed to compute a risk.
So the only solution is the SafetyNex solution : raw data are used locally to compute the risk, on the microprocessor of the smarphone, and those raw data are
NOT recorded in the cloud. Only risk statistics are recorded.


This technology differentiation allows SafetyNex to be the ONLY system that respects legal restriction to data recording (like in France for instance) and also the rules

of elementary ethics, and the proper respect for the privacy of drivers (even without law considerations, spying drivers does not match values of ​​NEXYAD).





Note: the time time latency can be guaranteed by NEXYAD because the computings are performed locally. Indeed, an App that would send and read data on the cloud could not guarantee latency
(it would depend on the network connection bandwith, and this is very variable). SafetyNex is then also the ONLY risk assessment App that is a real time application made to help the driver
while driving (it is not only an App for the Insurance Company).


As soon as the driver approaches an infrastructure with a speed showing that he/she did not understand the difficulty of the road, then SafetyNex warns a few seconds before danger in
order to let the driver slow down. It may save sriver’s life !


SafetyNex is not just a data collection tool for the insurer, it is also a useful tool for the driver, likely to save his/her life and at least to avoid causing accidents.


So using SafetyNex is a win win process : valuable for both insurer AND driver !


Read more :