Newsletter #22 is now available

NEXYAD Automotive & Transportation Newsletter #22, April 17th, 2018

 


4 disruptive AI algorithms for automotive mobility by NEXYAD


Headlines :

– SafetyNex episode 4 : Driving Risk Assessment for Automotive (Driving Assistant, ADAS, Autonomous Driving)
– CNEJITA Seminar on Artificial Intelligence: who will be responsible ?
– SafetyNex : driving robot maybe will mitigate human errors, but first they have to imitate good drivers
– “Theory of Water Flush” and Impact on the Prevention of Accidents for Autonomous Vehicles
– 4 disruptive AI algorithms for automotive mobility

Go to the Nexyad Automotive & Transportation Newsletter #22

4 disruptive AI algorithms for automotive mobility by NEXYAD

NEXYAD Automotive & Transportation Newsletter #22, April 17th, 2018

 


4 disruptive AI algorithms for automotive mobility by NEXYAD

Headlines :
– SafetyNex episode 4 : Driving Risk Assessment for Automotive (Driving Assistant, ADAS, Autonomous Driving)

– CNEJITA Seminar on Artificial Intelligence: who will be responsible ?

– SafetyNex : driving robot maybe will mitigate human errors, but first they have to imitate good drivers

– “Theory of Water Flush” and Impact on the Prevention of Accidents for Autonomous Vehicles

– 4 disruptive AI algorithms for automotive mobility

* * * * *



SafetyNex episode 4:
Driving Risk Assessment for Automotive
(Driving Assistant, ADAS, Autonomous Driving)

 
The new video on SafetyNex, on board driving risk assessment in real time.



* * * * *



CNEJITA Seminar on Artificial Intelligence:
who will be responsible ?

 
April 10th, CNEJITA (National Company of Legal Experts on Computer Science and Associated Techniques) organized a Seminar, whose objective is to determine the responsibility in terms of artificial intelligence through the understanding of technology and the dialogue with the actors of the sector.
It is therefore around this theme of topicality and future which is the artificial intelligence that the best experts in terms of computing met at the Commercial Court of Paris.

AI: concepts, technological breakthroughs and new risks
– Understanding the Concepts and Landscape of AI – Jean-Claude HEUDIN (Artificial-Creature.com – Teacher Researcher in AI)
– IA: state of play and perspectives – Jean-Philippe DESBIOLLES (IBM head of France IA WATSON)

Roundtable – Which Expertise fo AI ? was animated by Serge MIGAYRON (Honorary President of CNEJITA)
– The acceptability and limits of IA – JA CAUSSE (CNEJITA Expert)
– The Autonomous Vehicle and Traceability of IA – Jean-Louis LEQUEUX (Former President of VeDeCoM Tech)
– Auditability and risk control in the design of an IA – Gérard YAHIAOUI (NEXYAD)
– Evolution of the world of insurance, towards an objective responsibility – Nicolas HELENON (Co-manager Firm NEO TECH Assurances)

Roundtable – The Legal Challenges of AI. Animation – A MEILLASSOUX (ATM Lawyers – President of AFDIT)

– Introduction to Classical and New AI Concepts by Law: Applicable Regime and Evidence – L SZUSKIN (BAKER McKENZIE Lawyer)
– Tort liability in the face of AI: adaptation of traditional categories or creation of a responsibility specific to AI? – P GLASER (Lawyer TAYLOR WESSING)
– Contractual liability in the face of the IA: risk management during the contractualization of an IA system – FP LANI (DERRIENNIC Associate Lawyer)
– Synthesis on the current legal landscape – G de MONTEYNARD (Attorney General at the Court of Cassation)

Colloque CNEJITA
Gérard YAHIAOUI, CEO of NEXYAD

* * * * *



SafetyNex : driving robot maybe will mitigate human errors,
but first they have to imitate good drivers

 
BEWARE with the statistics : “94% of severe personal damage accidents are due to human errors” doesn’t mean that you’ll save 94% of severe accident with autonomous driving : drivers do not only make mistakes they also drive well (1 accident every 70 000 km, 3 dead every billion km – OCDE) … It is important to study also good driving and near misses (when driver has the right behaviour to avoid accident or to mitigate severity)… That’s what NEXYAD did during 15 years of research programs on road safety ^^ (that led to SafetyNex). See image (if you do not provide the “green” features, you will lose lives more than you gain with your driverless car. Our AI algorithm SafetyNex was made for this.

Example of risk tree

* * * * *



“Theory of Water Flush” and Impact on the Prevention of Accidents for Autonomous Vehicles

 
“THEORY OF WATER FLUSH” AND IMPACT ON THE PREVENTION OF ACCIDENTS
FOR AUTONOMOUS VEHICLES

by NEXYAD

INTRODUCTION
Let’s suppose that the flush does not exist in our toilets, and then let’s suppose that engineers able to create complex systems or even “systems of systems” are consulted to invent it, and that they apply exactly the same method than they do in the field of ADAS and Autonomous Vehicles.

Water Flush Vs Automotive Engineers

METHOD OF SCENARIOS
We propose to apply the method of scenarios, which consists in crossing all the factors that can modify the situation, then in each case of the complete combination, propose a solution. For this, it is necessary to note the number of possible shapes for the tank, the possible volumes, all the possible locations for the water supply entry, the possible diameters of the inlet pipe, the flow rates and possible pressures of water, the possible residual water levels before filling. We can generate the combinatorial of these factors, which allows us to generate all the possible scenarios of the “flush” problem. In each case, it is possible to give a solution, namely, the duration of filling of the tank (opening and closing of the water tap).

This approach is fully compatible with deep learning, which will also interpolate between two reference cases (quality of interpolation/generalization to be controlled, of course) if characteristics had to drift over time. Of course, the tank must integrate a system of sensors to evaluate the configuration (diameter of pipe, pressure of water, position pipe, capacity of the tank, etc …). We can use a camera, lasers, ultrasounds, etc. So that this recognition of situation is as accurate as possible. For such an approach, automation/control engineers talk about open-loop (feed forward) control because the data flow is as follows:

Open Loop

COST AND ROBUSTNESS OF THE SCENARIOS METHOD
It is easy to understand that the flush thus designed will be perfectly functional (there is no reason for it does not work), but for a high cost due to the sensors to integrate. Similarly, the robustness of the system to a measurement error or to a bad situation recognition is not guaranteed : we can very good to fill too much or not enough. The accuracy of the configuration case recognition is very important.


SOLUTION OF WATER FLUSH IN THE REAL WORLD
If you have the curiosity to disassemble your flush, you will notice that it is much simpler than the system described above: A float indicates when the water supply valve should be closed. The figure is as follows:

Closed Loop

Automation engineers call this a closed loop control (servo control). The feed forward “open” control is reduced to “open the tap thoroughly without worrying about the flow of water, the volume of the tank, and turn off the tap as soon as the float asks for it “. Note that this method works regardless of the configuration of the flush : we do not even need to know the volume of the tank that can be modified (for example: by filling half of the tank with glass beads) without affecting the operation of the flush. It is a robust and cheap system.


TRANSCRIPT OF THESE REMARKS IN THE FIELD OF ADAS AND AUTONOMOUS VEHICLES:
SERVO CONTROL IN DECISION

The information processing chain of the autonomous vehicle follows the general feed forward form :
NEXYAD has developed the SafetyNex system which dynamically estimates in real time the risk that the driver (human or artificial) takes. However, the autonomous vehicle may be functionally specified as follows:

AD feeding

“transport someone from point A to point B as quickly as possible, and safely.”

The “quickly” aspect is the historical business of the automobile. The “safely” notion integrates intrinsic safety of the system (its dependability: it should not explode, sensors or power supply may not be disabled, etc.), and since it is a vehicle, its ability to move with a good road safety, that is to say by “not taking too much risk in driving”. Since SafetyNex estimates this driving risk dynamically and in real time, it can be said that SafetyNex is a dynamic indicator of “SOTIF” (Safety Of The Intended Function). SafetyNex acts as a “driving risk float” : when the risk arrives at the maximum accepted level (like the float of the flush) we stop the action that raised the risk (example: we stop accelerating or we slow down). Thus, the response of an autonomous driving system is made adaptive (at the decision level) : even if the feed forward open loop is not perfect, it can correct itself to take into account, among other things, the instruction and the measure of driving risk. This system is completely independent of the automatic driving system in terms of information processing, so it represents redundancy of processing.

SafetyNex uses to estimate risk :

. risk due to inadequacy of driving behaviour to the difficulties of the infrastructure : navigation map, GPS, accelerometers

. risk due to inadequacy of driving behaviour to the presence of other road users (cars, pedestrians, …) : data extracted from the sensors (camera, lidar, radar, etc) such as “time to collision”, “inter distance (in seconds)”, number of vulnerables around, etc.

. risk due to inadequacy of driving behaviour to weather conditions: in particular to atmospheric visibility (fog, rain, snow, sand, penumbra). Knowing that when visibility is low, vehicle must pay more attention (and slow down) even if this autonomous vehicle is not impacted by the decrease in visibility (if it only uses a lidar for example) because the avoidance of an accident is done at the same time by the two protagonists : if one of them (pedestrian, human driver), does not see the autonomous vehicle, then it finds itself only to be able to avoid the accident, which doubles the probabilities of a potential accident.

. other

The use of SafetyNex allows to make adaptive an artificial intelligence of autonomous driving, on the following diagram :

Adapt Closed loop AI

If you have a lean computer, then you only apply one loop between t and (t+1) as it is shown on the figure. If you have a powerful computer, you can then even simulate a big number of decisions and take the less risky one (like automaticians do with predictive control systems). Of course, SafetyNex is only ONE way to close the loop (on a crucial notion : driving risk). This figure may be extanded to other variables of contol that make sense for an autonomous vehicle. More complex adaptation rules may switch from a decision to another if risk simulation shows that finally it is less risky (ex : slow down or turn wheel ?).


CONCLUSION
SafetyNex uses the map in addition to sensors (same sensors as the driving system or parallel tracks) and does not need to accurately identify the situation but instead to estimate a risk (this is a different task). SafetyNex is a knowledge-based AI system (knowledge extracted from human experts in road safety, from 19 countries – Europe Japan USA – who validated the system over 50 million km. Total research program duration : 15 years). This technology is still being improved, of course, but it can already be integrated into autonomous vehicles and avoid a large number of accidents by its ability to make the system adaptive to unknown situations. In particular, in the case of autonomous urban vehicles (autonomous shuttles, robot taxis), the adaptation of driving behaviour to complexity of infrastructure is made possible by SafetyNex, which decodes this complexity by reading the navigation map in front of the vehicle. SafetyNex makes the autonomous vehicle anticipate more by following “rules of safety” : with SafetyNex emergency situations (that still will need emergency braking and other emergency actions) become much more rare. Autonomous vehicle acts like an experienced cautious driver. Note : if you modulate Maximum Accepted Risk, then you modulate aggressiveness of the autonomous vehicle. This might make sense not to let the autonomous vehicle trapped in complex human driving situations (where the autonomous vehicle would stopped indefinitly).

* * * * *



4 disruptive AI algorithms for automotive mobility

 

. ObstaNex detects obstacles with a simple cam (a la Mobileye).
What is disruptive ?
ObstaNex runs in real time on a regular smartphone… it means it doesn’t need a big computing power to run. It can be trained/re trained on a “small database” using the methodology A.G.E.N.D.A. (Approche Générale des Etudes Neuronales pour le Développement d’Applications or General Approach of Neuronal Studies for Application Development) – important is you improve your cam !
 



. RoadNex detects drivable part of the lane borders and free space.
What is disruptive ?
RoadNex works even in the Streets of old cities as Paris, London or Roma, and it runs in real time on a regular smartphone. it means it doesn’t need a big computing power to run.
 



. VisiNex detects lacks of visibility (fog, heavy rain, snow, sand storm …).
What is disruptive ?
VisiNex is an artificial vision tool which is correlated with human perception. If there is something to see, VisiNex is able to give a score of visibility. Except Daimler, we haven’t seen such a military background-based detection elsewhere.



. SafetyNex is the only fusion Artificial Intelligence algorithm (sensor + map fusion) that estimates driving risk dynamically and in real time.
What is disruptive?
SafetyNex allows to have an explicit value of driving risk. It is a total revolution for car insurers, fleet managers, and autonomous driving engineers. These algorithms are already under integration into products for telematics /connected car, ADAS, Autonomous Vehicle.


SafetyNex : driving robot maybe will mitigate human errors,
but first they have to imitate good drivers

BEWARE with the statistics : “94% of severe personal damage accidents are due to human errors” doesn’t mean that you’ll save 94% of severe accident with autonomous driving : drivers do not only make mistakes they also drive well (1 accident every 70 000 km, 3 dead every billion km – OCDE) … It is important to study also good driving and near misses (when driver has the right behaviour to avoid accident or to mitigate severity)… That’s what NEXYAD did during 15 years of research programs on road safety ^^ (that led to SafetyNex). See image (if you do not provide the “green” features, you will lose lives more than you gain with your driverless car. Our AI algorithm SafetyNex was made for this.

Example of risk tree

Interpretation of Risk Profiles with SafetyNex
A new paper by Nexyad

ESTIMATION OF DRIVER’S EXPERTISE, OF RISK TAKEN CONSCIOUSLY BY THE DRIVER,
AND OF THE LACK OF ANTICIPATION AND MISUNDERSTANDING OF THE ROAD



By NEXYAD

Lire la version Française


I – INTRODUCTION
NEXYAD has been developing the smartphonte application SafetyNex which estimates the risk of driving in real time [1]. SafetyNex is both a driver assistance system (ADAS), which alerts the driver (vocal alert) before danger (When the risk increases too much), and a telematics system that records risk profiles and usage profiles.
Warning before the danger gives the driver time to slow down and avoid the accident. Road Safety studies show that SafetyNex can reduce the number of accidents by 20% [2]. This simple functionality is of interest of car insurers, fleet managers, and to car manufacturers.
SafetyNex also rewards the driver with cups (gold, silver, bronze) that can be transformed into money incentive (vouchers, etc.) so that the safe drivers stil have a daily interest to go on using SafetyNex. Indeed, tools that are not used over a period of time rarely have a real effect on the accidentology. SafetyNex is therefore distinguished from other products, on the one hand by its real time and driving assistance, but also for its “reward” side. SafetyNex informs the driver In real time when the risk exceeds a threshold of danger, than one can say that SafetyNex gives the risk in the hands of the driver first. The driver is in control of his/her risk.
Then SafetyNex distinguishes from all telematics products that ultimately provide information to the insurer or fleet manager, but not to the driver who feels rightly spied on.
Risk and usage profiles [3] are forwarded to managers who have an interest in minimizing risk and the number of accidents. This paper presents a simple way to interpret the risk profiles constructed by SafetyNex.

II – SafetyNex RISK PROFILES
SafetyNex estimates risk of driving at every instant.
Since SafetyNex also measures usages, it measures among other things the durations and the number of traveled kilometers.
One can then construct the curve Risk (t) which is the risk at each moment, and also the curve Risk (km) which is the risk at each point of the route.

Risk Distance

Risk Distance rated

Let’s consider one or the other of those curves, it is easy then to cut the risk into slots:

Risk 03

It is therefore possible to calculate the total duration spent [resp the total number of km carried out] with a risk
between 10% and 20%, for example (or between 50% and 60%). The graph of these durations [resp number of km] for each range of risk (0% -10%, 10% -20%, etc.), looks like :

Graph 4

It can be seen that this graph can be seperated into three parts:

Graph 05

. A very high bar of near zero risk
. A shape of « bell curve » comparable to a gaussian
. Rising at the very end towards high risks
NEXYAD has run over 3,500 testers since June 2016, and has been able to interpret the shapes of these curves.

III – INTERPRETATION OF SafetyNex RISK PROFILES : CONSENTED RISK, EXPERTISE OF DRIVING, LACK OF ANTICIPATION OF DRIVER
The large quasi-zero risk bar simply expresses the fact that overall the car is a safe mode of transportation.
The part that draws a bell curve has a more or less strong spread : we have noticed that experienced drivers have a narrow curve (repeatability of their driving style is high) while beginners have a huge spread (they can’t drive always the same way).
The centering of the bell curve (maximum likelihood) corresponds to the way in which the driver takes a controlled Risk : cautious beginners have a low maximum likelihood (they try to take as less risk as they can) while experienced drivers have a higher maximum likelihood : they know what risk level thay can cope with.
Finally, the values that go up to the right (tail of distribution of the curve in bell) correspond to the vocal alerts, that is to saycases where the driver has not fully understood that the risk is high. In other words, it is the lack of anticipation and misunderstanding of road.

IV – CONCLUSION
SafetyNex’s risk profiles make it possible to understand the kind of driver you have :
. Cautious / not cautious (maximum likelihood of the bell curve position)
. Experienced / beginner (spreading of the bell curve)
. Lack of anticipation / very good anticipation (queues of distribution of the curve in bell)
Fleet insurers and managers therefore have all the information that they need to help the driver.
For example, within the framework of prevention plans, offering training adapted to each type of risky driving.
We validated this information by driving 3,500 testers, including beginners, experienced drivers, and also pilots (who in take risks appearance, but in reality have a very safe driving). This allowed us to give these interpretations of SafetyNex’s risk profiles.
With the deployment of SafetyNex to reduce the number of accidents, professionals structurally gain margin, and can use this margin to analyze profiles, segment them, and find the segments where it may be interesting to develop UBI (Usage Based Insurance) and real time pricing fleet.
This multi-functionality of SafetyNex makes it a unique and effective tool for managing driving risks.

V – REFERENCES
[1] : THE ULTIMATE SOLUTION FOR INSURANCE COMPANIES THAT NEED ONBOARD RISK ASSESSMENT : SafetyNex

[2] : SMARTPHONE APP SafetyNex COULD REDUCE ACCIDENT RATE BY 20%

[3] : EXAMPLE OF SafetyNex RISK PROFILES