Artificial Intelligence and Road Safety

NEXYAD Automotive & Transportation Newsletter #30, September 14th, 2020


Artificial Intelligence and Road Safety


Headlines :

– Nexyad CEO Editorial

– Why do we develop real time hybrid Artificial Intelligence for road safety

– OTN and Nexyad join forces in a strategic partnership for active rail safety

– SafetyNex Integrator BRIGHTMILE unveils partnership with AXA Insurance

– RoadNex Off Road Drivable Surface detection

– Hazard, Criticality, and Risk in Road Safety ?

– Gathering Computer Vision Detection with Prediction and Risk Assessment: This is what we do!

– SafetyNex: ADAS for Road Safety Using Map


* * * * *


Nexyad CEO Editorial



The world is going through an unprecedented crisis and more than ever it is innovation that will allow industries to take advantage of the opportunities created by destabilization of markets. In this context, it is interesting to notice that NEXYAD finds itself at the center of many strategies of automotive industrial groups, rail, and mobility.

NEXYAD’s embedded software modules are intended to be integrated into application software programs and embedded electronic application platforms, and we are now present on several markets:

  •  advanced telematics, which deploys smartphone apps, electronic devices, and dashcams
  •  automotive, which deploys advanced driver assistance systems, and connected vehicle features
  •  public transport, including bus, tram, metro, train, etc.
  •  advanced autonomous vehicle projects
  •  road safety for motorcycles and bicycles

NEXYAD provides computer vision functions:

  • detection of any lack of visibility, with the VisiNex module (lack of visibility is a factor of risk,
    VisiNex detects fog, heavy rain, sand storm, snow storm, lack of lighting, dust on embedded cameras, …)
  • drivable surface detection, with the RoadNex module
  • detection and recognition of objects, obstacles, ‘a la MOBILEYE’, with the ObstaNex module

Note: VisiNex may also be used to automatically adapt parameters of a complex computer vision chain to image quality. This is used by NEXYAD to increase number of use cases where our computer vision modules lead to good performance. And when image quality (measured with SafetyNex) is not enough for any reliable detection, then NEXYAD modules are aware of it, they can end a special output “not applicable” which is very important for sensor fusion stages.

And our  company also strongly deploys its unique real time driving risk assessment* module SafetyNex, validated in terms of metrics, by global insurers, police of road safety,  one autonomous vehicles manufacturer, and major car OEMs. The current and upcoming deployments in the very short term are in Europe, USA, Asia, India, and China. We are proud of the deployments planned by our integrator customers, because it will allow us, through their high-level products, to save lives, which is our mission statement.

SafetyNex is a Knowledge Based System (Hybrid Artificial Intelligence) involving symbolic reasoning) that monitors in real time driving behavior to detect 20 time per second any lack of cautious. This helps not to go to emergency situations, and if ever, with a driving behavior (vehicle speed and accel) that will be adequate to let more time to ADAS systems for their detection, criticality assessment, decision-making, and action.

This is what we call ANTICIPATION skills.

Real time applications of SafetyNex today are:

  • alerting driver through advanced telematics tools (smartphone App, telematics device, dashcam), for fleets and insurers risk management
  • alerting driver through advanced cockpit HMI systems
  • automatically modulating vehicle speed and accel of ACC (Automated Cruise Control)
  • alerting the AI od AD system that risk is rising

*_Risk Assessment shouldn’t be confused with hazard detection or with criticality assessment during emergency situations. Risk is lack of cautious, it is NOT probability of collision,  and it must be assessed permanently during normal driving. (see our article on risk assessment  in the following articles).

We Save Lives


* * * * *

Why do we develop real time hybrid Artificial Intelligence for road safety


« We believe that every single Human life on Earth deserves to be saved. We can avoid up to 840 million road accidents and save up to 250 thousand lives per year with Minimum Viable Version of our technology. Road accident is a huge factor of material & resources waste (repair) and pollution/CO2 emissions: by saving lives we also contribute saving the planet. After more then 25 years of high level R&D, we propose today validated onboard AI technology deployed by our customers through easy to use products, for a better road safety »

Check our 1 minute global presentation video :




* * * * *


OTN and Nexyad join forces in a strategic partnership for active rail safety


The companies OTN and Nexyad join forces in a strategic partnership for active rail safety.

Saint Germain en Laye, April 20, 2020 – The company OTN, which develops safety products in the rail sector among others, announces the signature of a partnership with the company Nexyad, specializing in on-board risk assessment of accident, with the aim of developing solutions for predicting and preventing collision risks.

This partnership will significantly improve the performance of current detection systems, by offering a risk detection horizon adapted to the speed of traffic on trams, metros and trains.

This technological contribution will allow rapid migration from simple “detection” to “prediction”, thereby improving the overall safety of users and the environment close to the traffic lanes.

The combination of the skills and experience of our two companies will allow us to quickly develop a high-performance analysis system (images, signals, data) to identify the risks of railway equipment in traffic. After six months spent evaluating the work necessary for the development and adaptation of the algorithms, initially intended for road safety, we are happy that this partnership results in the development of prototypes which we will be able to test in real situation in the short term.

OTN, a company based in Saône et Loire, France, manufactures, designs and markets electrical and electromechanical equipment intended for various advanced markets, including railway.

Nexyad, a company based near Paris, markets a global road safety platform for telematics, Advanced Driver Assistance Systems, and Autonomous Vehicles, currently adapted to railway focus.



* * * * *

SafetyNex Integrator BRIGHTMILE unveils partnership with AXA Insurance


“AXA Insurance is offering the Brightmile app to its fleet customers as a way of supporting safe driving during the COVID-19 pandemic.

The motor insurer and the tech start-up are partnering on a smartphone telematics solution for fleets, which they are launching during lockdown.”

To reed more:




* * * * *

RoadNex Off Road Drivable Surface Detection




In this demo video, NEXYAD RoadNex algorithmes are challenged with very bad conditions : There is no wheel angle, off road with kind of white gravels, high light thru trees in the darkness of wood which bring lots of shadows on the trail. The result is pretty good the green highlighting shows where the vehicle is able to go!



* * * * *

Hazard, Criticality and Risk in Road Safety


Meanings of those words are very different and it is very important to distinguish them:

. HAZARD indicates presence of POTENTIAL DANGER: if behind a door there are toxic materials, you write “HAZARD” on the door. But if the operator who opens the door has the appropriate helmet, he/she can do so at no RISK.

. CRITICALITY indicates a measurable level of DANGER. During a road accident, CRITICALITY is measured by “time to collision”, “probability of interception of trajectories”, etc.

. RISK indicates the absence of caution: if you put a glass near the edge of the table, you take a RISK. In this case HAZARD is the edge of the table. If the glass falls, at every moment you can measure CRITICALITY of the situation. If you have good reflexes, you can catch the glass. This is what AEB does for vehicles (ADAS). If you want to avoid breaking the glass without using reflexes, you put the glass away from the edge. In road safety, it’s the same, and NEXYAD has developed a real-time embedded software tool that measures “how far we are from the edge”. Applications are: alert driver, or act automatically on the driving behavior of the vehicle (ACC, Autonomous Driving). thereby improving the overall safety of users and the environment close to the traffic lanes.





* * * * *

Gathering Computer Vision Detection with Prediction and Risk Assessment:
This is what we do!


Because most automotive engineers have no knowledge in Road Safety, they usually do not know that 75% of driving-risk come from a driving behaviour that is inadequate to infrastructure characteristics (geometry of curve, narrowness, pattern of intersection – angles, number of input and output lanes, functionality of infrastructure – pedestrian crossing, school zone, … -, road signs – traffic light, stop sign, …). So they focus on vehicle performance in passive safety (deformation of vehicle during a crash, safety belt, air bag, etc… and active safety, vehicle detection and reflexe : emergency braking (AEB). NEXYAD do not build cars, but as software company, we worked for 20 years with road safety experts and road infrastructure experts on 8 major scientific programs to acquire this specific knowledge. This is why our minimum viable product for road safety can run onboard on a smartphone or any other device, and reduces accident rate by 25% with only a map, accelerometers, and GNSS. This year we can add NEXYAD computer vision modules that bring other risk factors; ObstaNex which brings interdistance and presence of vulrenables, RoadNex which brings size of drivable surface, and VisiNex which detects lacks of visibility. This brings a complete onboard road safety platform: under deployment for telematics (fleets and insurers), ADAS, and Autonomous Vehicle. Available for cars, motorcycles, buses, trucks, trams, and soon for bicycles and other vulnerable mobility systems. It is time to join modernity, and integrate our best technology into your valuable solutions, available everywhere…  





* * * * *

SafetyNex: ADAS for Road Safety Using Map


Sometimes, accurate detection both by human perception and by sensors are not enough for road safety. Nexyad has worked on some use cases for which classical ADAS can’t help. So we have developed SafetyNex that reads and analyses the digital map (HERE, TOMTOM, OSM, etc.) in front of the vehicle
(4-5 seconds ahead) as a bird view to anticipate singularities of the road. Then SafetyNex allows to vocally alert drivers when they approach these singularities (curves, intersections, pedestrian crossing, etc…) in order, for them, to reduce speed and to lower risk of accident.   Of course, it is possible to equip vehicles with Adaptative Cruise Control (ACC) to reduce speed automatically before these singularities. As a reminder, road safety studies show that if speed is reduced of 1mph to the right moment, risk of accident is lowered of 4% to 6%. With SafetyNex alerts, drivers have time (at least 4 seconds) to reduce speed far more than 1mph.


See 2 use cases below :


* * * * *