New Release of RoadNex v2.3 available in RT-MAPS 64bits
The new release of NEXYAD Road Detection RoadNex v2.3 is now available in RT-Maps 64bits (by Intempora).
RoadNex detecting track in desert
The new release of NEXYAD Road Detection RoadNex v2.3 is now available in RT-Maps 64bits (by Intempora).
RoadNex detecting track in desert
New Release of Road Detection Nexyad RoadNex v2.3 (32bits) is twice Faster.
Road Detection in Traffic
Nexyad Team validated that RoadNex v2.2 works in real time on smartphone !
Soon available on IOS, Android and Windows Phone.
Demo of NEXYAD technology for ADAS.
Nexyad combines one camera with another one for obstacle detections (ObstaNex v2.1) and uses data fusion to reshape stereo.
This solution still can work in mono when one camera is off.
In this example, we placed cameras in the place of rearview Mirror on a van.
Detection of pedestrian near the vehicle
Early Metrics, first rating agency for startups gave three stars to NEXYAD.
The audit responded to an external demand by a large industrial company. The founders, the project, the market and the financial statement of NEXYAD were scanned by experts for a score of 75/100.
More informations : Early Metrics
« Handpicked pioneering executives from the most innovative insurers, brokers and automakers leading the way in insurance telematics. See a unique collection of speakers to be found at no other event. »
The NEXYAD Team will present SafetyNex.
Speakers’ List & Bio here : http://www.tu-auto.com/connectedcar-insurance-eu/conference-speakers.php
DEDICATED TO THE TESTING, DEVELOPMENT, AND VALIDATION OF AUTONOMOUS VEHICLES AND NEXT GENERATION ADVANCED DRIVER ASSISTANCE SYSTEMS!
31 MAY – 2 JUNE 2016 | STUTTGART, GERMANY
Full list of speakers confirmed include:
• Daniel Benhammou, CEO, Acyclica Inc
• Wolfgang Herzner, senior engineer, AIT Austrian Institute of Technology GmbH
• Paul Krutko, President & CEO, Ann Arbor SPARK
• Sandeep Sovani, director, global automotive industry, Ansys
• Marcos Pillado, project manager, Applus+ IDIADA
• Henning Lategahn, managing director, Atlatec GmbH
• Jeffrey Ferrin, head of research, Autonomous Solutions Inc
• Juergen Holzinger, project manager AVL-Drive, AVL List GmbH
• Alexander Noack, head of sales, b-plus GmbH
• Philipp Kerschbaum, HMI development, BMW Group
• Igor Doric, scientific and technical manager, CARISSMA
• Stefan Lüke, project manager, Continental Division Chassis & Safety
• Reija Viinanen, managing director, Fell Lapland Business Services
• David LaRue, manager ADAS/AD systems, FEV
• Francesco Crisci, senior test engineer, Fiat Chrysler Automobiles
• Adrian Zlocki, head ADAS department, FKA
• Manuel Merz, research engineer, Ford
• Stefan Wolter, HMI specialist, Ford
• Frederik Diederichs, senior researcher, Fraunhofer IAO
• Arno Eichberger, associate professor, Graz University of Technology
• Tom Lueders, director tools and testing, Hella Aglaia Mobile Vision GmbH
• Chris Reeves, commercial manager, future transport technologies, Horiba Mira Ltd
• Alexander Treis, business development manager, IEE SA
• Alice Siu Man Chan, senior research engineer, Institute for Infocomm Research
• Fu Keong Chia, research engineer, Institute for Infocomm Research (I2R), A*STAR
• Peter Vertal, head of research, Institute of Forensic Engineering, University of Zilina
• Fabian Bauer, engineer software development, IPG Automotive GmbH
• Edwin Nas, deputy project leader self-driving vehicles, Netherlands, Ministry of Infrastructure & the Environment
• Lyn Matten, managing consultant, MM1 Consulting & Management PartG
• Volker Scholz, managing partner, MM1 Consulting & Management PartG
• Dominic Gallello, CEO, MSC Software
• Gérard Yahiaoui, President and CEO, Nexyad
• Rien van der Knaap, managing partner, OC Mobility
• Mugur Tatar, managing director, QTronic GmbH
• Gerben Feddes, senior advisor intelligent mobility, RDW
• Heather Stoner, division manager, Realtime Technologies Inc
• Arnd Engeln, professor of Market and Advertising Research, Traffic and Transport Psychology, Stuttgart Media University
• Robert Friis, president, Summit Development Group LLC
• Saskia de Craen, senior researcher, SWOV Institute for Road Safety Research
• Rikke Kuipers, senior security specialist, Synopsys
• Brian Ceccarelli, owner, Talus Software
• Robin van der Made, product manager software and services, TASS International
• Padmanaban Dheenadhayalan, engineer, Tata Elxsi Ltd
• Maria Kreußlein, research assistant, Technische Universität Chemnitz
• Nicholas Clay, senior manager – testing, Thatcham Research
• Andrew Miller, chief technical officer, Thatcham Research
• Rebecca Advani, senior technologist, Transport Systems Catapult
• Nick Reed, academy director, TRL
• Paul Newman, BP Professor of information engineering, University of Oxford
• Alain Piperno, autonomous vehicles testing/homologation project manager, UTAC CERAM
• Patrice Reilhac, innovation and collaboration research director, Valeo CDA
• Oscar Slotosch, member of the board, Validas AG
• Ingo Nickles, field application engineer, Vector Software
• Diego Minen, technical director, VI-Grade
• Carina Björnsson, technical expert, driver assistance and active safety test methods, Volvo Car Corporation
• Aki Lumiaho, principal scientist, VTT Technical Research Centre of Finland Ltd
• Christian Purucker, project manager, WIVW GmbH
• Gunwant Dhadyalla, principal engineer, WMG, University of Warwick
Automakers intending to bring driverless cars to market need to work as much on software design as mechanical engineering, the researcher leading Nissan Motor Co.’s automated-vehicle program said.
Making cars that are “deliberative” in assessing road conditions, rather than just reactive, requires artificial intelligence, Maarten Sierhuis, director of Nissan’s Silicon Valley research center in Sunnyvale, California, said in an interview. The carmaker, which aims to sell vehicles that can drive themselves by 2020 or sooner, is developing software to read and filter sensor data much as a human brain does, he said.
To read more : http://www.newsday.com/classifieds/cars/experts-say-driverless-cars-will-require-artificial-intelligence-1.8817975
NEXYAD has been working on such an Artificial Intelligence through their risk of accident computing module SafetyNex.
SafetyNex is a knowledge-based system that computes a huge data fusion : information about the navigation map, speed of the car, accelerations, visibility, grip(adherence), time to collision, interdistance, … This data fusion aims to compute a risk of accident : low score means the driver (human driver … or Artificial Intelligence, robot) drives in a safe way. This SafetyNex is an artificial intelligence that should work embeded with the driving robot.
The first market of SafetyNex is Insurance companies that need to score the behaviour of drivers (pay how you drive …), it doesn’t need to be as precised and reliable than it should be for a driverless car, but NEXYAD started to deploy and then will be the most experienced ADAS/AI company on this subject in 2020 when autonomous cars will invade the streets.
Nexyad made experimentation in the same path with four different driving behaviour, each second we recorded points that represent a score of safety and of Eco driving. The more high on the graphic is the points the more safe is the driving, the more right the more Eco :
18 Feb 2016 :
NEXYAD CEO Gerard YAHIAOUI, and a panel of SMEs founders invited by Isabelle STHEMER of DojoGroup, in Paris.
The main purpose was to share ideas that may enhance SMEs performance and French competitivity.
Gérard Yahiaoui in a very casual discussion.
The Senator Claude NOUGEIN (here) that dares to talk to the base for a better understanding of the micro economy in the country. Quite unusual in France and very important !
Thank you Monsieur le Senateur !
NEXYAD explique pour les Assureurs que les solutions de télématique actuelles ne permettent pas et ne permettront jamais de calculer un risque d’accident. NEXYAD présente une innovation qui propose une approche totalement disruptive du calcul de risque : SafetyNex
Lire l’article Calcul du risque d’accident pris par les conducteurs. Fév 2016 (pdf)
According to the newspaper ‘Les Echos », AG Insurance gave up linking driving behavior to accident. For eight months, the insurer has equipped the car of staff members with « smartbox ». Verdict? Nah, nothing interesting linkable to driving style !
In 2014, the bank insurer KBC had reached the same conclusion after a test conducted with 150 employees, whose car was equipped with a device provided by TomTom. »
For Nexyad, there is no mystery. Until now the applications or cases expected to provide data on the behavior of drivers all operate on the same principles :
– severe acceleration and braking
– Vehicle geolocation
– Cornering speed / curves
– Usage (mileage, road types, driving schedules, for example)
This information does not measure the « good behavior » that may inform actuaries of insurance on real risks taken by their clients.
Indeed, the correlation between the brutal acceleration (or severe braking) and road safety is a mistaken belief. The assumption that severe braking would report to a lack of anticipation of the driver, and instead a slight braking would be a sign of good anticipation is definitely wrong. This has been formally shown and demonstrated by experts in road safety. Besides, these same experts believe that an Eco driving is necessarily dangerous: Eco driving mainly keeping as most as possible the vehicle’s inertia, and thus slowing down as little as possible, which would push the Eco drivers to delay their decision of braking including when they approach a vulnerable road user (pedestrians, bicycles, scooters, etc.).
Nexyad worked since 2001 on these issues through four national collaborative research programs and now has a totally disruptive application that measures the Eco driving, but also road safety and those two measurements has no correlation).
This work consisted, for 15 years, in building the causal rules that lead to accidents. This is entirely different of statistical approaches. An insurance actuary may find that the majority of drivers under 25 years old with a red car have more accidents than others, and then this actuary could modulate its pricing on this criterion. This is reasoning on correlations and not cause and effect. Nexyad is the first and only company that can bring cause and effect considerations into accident risk assesment for insurance companies.
Nexyad developed the module SafetyNex which decodes the difficulties of the road infrastructure (dangerous curves, intersections, pedestrian crossings, school zone, etc.) relates those difficulties to the behavior of the vehicle through a grid of cause and effect. The construction of this grid of cause and effect required 15 years of knowledge extraction among a panel of road safety experts (their job is to explain accidents, and find ways to reshape roads and infrastructures in order to reduce statistics of accidents – national and European).
SafetyNex has been validated by INRETS (now became IFFSTAR) that deployed 500 vehiclesand compared risk estimation by SafetyNex with the accidents database « MAIS » of the National French Gendarmerie. We could demonstrate that SafetyNex estimates risk of traffic accident, as an immediately usable score for actuaries of insurance (for example).
NB: Accidents are rare events (one every 70 000km on average). This is why teams of road safety improvement at the national level have set up « observatories of cars trajectories ». They allowed to count the « near misses ». A near miss is accident that were prevented at the very last second (because one of the drivers had the proper reflex, …). When observing a repeat of near misses, then you end up getting an accident (near misses are at least a hundred times more numerous than accidents).
Nexyad is the only team of telematics solutions that brings this knowledge of these near misses and that is able to share it with the risk estimation experts at insurance companies.
How it works ?
Every second, a risk score and an Eco score are computed by SafetyNex, depending on the vehicle speed relative to the difficulty of the road infrastructure. Drivers can be warned in advance, 4 seconds before arriving on the dangerous area. Then SafetyNex may also decrease the number of accident (or the severity of accidents when they can’t be avoided).
SafetyNex establishes daily or weekly reports on the risk and eco and shows what could have been better (advices on driving efficiency improvement).
Insurers may get reports too : the real risk indicators. This is what Nexyad called SafetyNex signature driving behavior (duration at every level of risk, for instance).
View demos 4 films on the same route (1mn30):
Example of a good driver : both safe and Eco.
Example of a bad driver : both waste and risky
Example of a sporting driver : waste but safe most of the time
Selon les Echos « AG Insurance renonce à lier la prime auto au comportement. Pendant huit mois, l’assureur a équipé la voiture de collaborateurs d’une « smartbox ». Verdict? Bof, pas de quoi lier la prime au style de conduite.
En 2014, le bancassureur KBC avait tiré les mêmes conclusions d’un test mené auprès de 150 collaborateurs, dont la voiture avait été équipée d’un dispositif fourni par TomTom. »
Lire la suite : http://www.lecho.be/entreprises/services_financiers_assurances/AG_Insurance_renonce_a_lier_la_prime_auto_au_comportement.9729053-3028.art?ckc=1&ts=1455208594
Pour Nexyad, il n’y a pas de mystère. Jusqu’à aujourd’hui les applications ou boîtiers censés fournir des données sur le comportement des conducteurs fonctionnent sur de mêmes principes falsifiables :
– remontée des données d’accélération et de freinage
– géolocalisation du véhicule
– vitesse dans les virages/courbes
– usage du conducteur (kilométrage, types de route, horaires de conduite par exemple)
Ces informations ne permettent pas de mesurer la « bonne conduite » et de renseigner les actuaires des assurances sur les risques réels pris par leurs clients.
En effet, la corrélation entre la brutalité des accélérations d’une part, la sévérité des freinages d’autre part et la sécurité routière est une croyance erronée. L’hypothèse selon laquelle un freinage sévère relèverait d’un manque d’anticipation du conducteur, et au contraire qu’un freinage léger serait le signe d’une bonne anticipation est fausse. Cela a été démontré formellement par les experts de la sécurité routière. D’ailleurs ces mêmes experts sont persuadés qu’une conduite Eco est nécessairement dangereuse : conduire Eco c’est garder l’inertie du véhicule, et donc freiner le moins possible, ce qui pousserait les conducteurs Eco à retarder leur décision de freinage y compris devant un vulnérable (piétons, vélos, scooters, etc.).
Nexyad a travaillé depuis 2001 sur ces questions au travers de 4 programmes de recherches nationaux collaboratifs et dispose aujourd’hui d’une application totalement disruptive qui mesure l’Eco conduite, mais aussi la sécurité routière.
Ce travail a consisté durant 15 ans à construire les règles de causalité des accidents. Cela est entièrement différent des approches statistiques. Un actuaire d’assurance qui constaterait que la majorité des conducteurs de moins de 25 ans possédant une voiture rouge ont plus d’accidents que les autres pourrait moduler sa tarification sur ce critère. Ceci est un raisonnement sur des corrélations et non pas sur des relations de cause à effet.
Nexyad a développé le module SafetyNex qui est à ce jour le seul au monde à décoder les difficultés de l’infrastructure routière (courbes dangereuses, intersections, passages piéton, zone écoles, etc.) rapportées au comportement du véhicule selon une grille de relations de cause à effet. La construction de cette grille a nécessité 15 ans de recueil d’expertises auprès des experts chargés de réaménager les routes afin de réduire les statistiques nationales et européennes d’accidentologie.
SafetyNex a été validé par les roulages de 500 véhicules de l’Institut de recherche INRETS devenu IFSTTAR, puis par des comparaisons d’estimation du risque avec la carte des accidents de la route du fichier « MAIS » de la gendarmerie nationale. Nous avons démontré que SafetyNex mesure un risque d’accident de la route, variable immédiatement utilisable pour les actuaires des assurances (par exemple).
NB : Les accidents sont rares (un tous les 70 000km en moyenne). C’est pourquoi les équipes chargées de la sécurité routière au niveau national ont constitué des observatoires de trajectoires qui permettent de compter les « quasi-accidents ». Un quasi-accident est un accident évité de justesse. Lorsqu’on observe une répétition de la fréquence de quasi-accidents, on finit par obtenir un accident (les quasi-accidents sont au minimum cent fois plus nombreux que les accidents).
Nexyad est la seule équipe proposant des solutions télématiques qui a connaissance de ces quasi-accidents, qui intègre les règles de causalité de l’accidentologie à son module SafetyNex et qui peut les partager avec les experts d’estimation de risque et des bases de sinistralité des assurances.
Comment ça marche ?
Chaque seconde une note de risque et une note Eco sont mesurées par SafetyNex en fonction de la vitesse du véhicule rapportée à la difficulté de l’infrastructure routière. Le conducteur peut être prévenu en avance (4 secondes) si le risque mesuré le met en danger lui et les autres usagers. SafetyNex établit des rapports journaliers ou hebdomadaires (par exemple) sur la conduite de l’utilisateur et les points qu’il devrait améliorer.
Les assureurs peuvent éventuellement disposer des rapports de conduite de leurs assurés et récompenser ceux dont la conduite est sûre par rapport aux vrais indicateurs de risque. C’est ce que Nexyad appelle la signature SafetyNex de comportement en conduite.
Voir 4 films démos sur un même parcours (1mn30) :
L’hebdomadaire Le Point de cette semaine sort un cahier spécial sur la ville de St Germain en Laye dans le département des Yvelines. La journaliste Valérie PEIFFER y a consacré une page à Nexyad.
« Née en 1995, Nexyad était une PME spécialisée dans la modélisation mathématique.
Reine des algorithmes, elle s’était imposée comme une société de recherche sous contrat qui travaillait
pour différents secteurs tels que l’armée, les banques, les assurances ou encore les centres d’épidémiologie. «Testés et éprouvés, nos modèles mathématiques de traitement de données, de signal et d’images tournent parfaitement pour analyser et mixer différents paramètres, offrir des solutions, valider les performances d’un produit ou encore mesurer l’efficacité d’une campagne de communication, explique Gérard Yahiaoui, cofondateur de Nexyad. Mais quand, en 2013, il devint possible de faire tourner nos modèles dans un simple smartphone, cela nous a ouvert de nouveaux horizons de recherche avec à la clé de nouveaux produits. »
STUTTGART 31 may – 2 june 2016
june 1st – 09:30 – Building a relevant validation database for camera-based ADAS
Validation of camera-based artificial vision systems applied on open world is a very complex issue. An HD colour camera may generate more than 65 000 power 2 000 000 different images (information theory), so it is not possible to test every possible message. We propose a deterministic approach for building a validation database using the AGENDA methodology that was developed and published in the 1990s for neural network database (learn and test) design.
NEXYAD is proud to announce that our project SEMACOR has been approved.
This project will help to us to accelerate applications of road safety/ risk of accident estimation
in embedded connected devices (based on the NEXYAD software SafetyNex) : new vehicles and also telematics devices (aftermarket applications for car industry, insurance companies, etc …)
Contact NEXYAD if you need to measure both safety and eco-driving onboard : (sales@nexyad.net)
Demo of SafetyNex : https://nexyad.net/Automotive-Transportation/?page_id=441
NB : SafetyNex is THE ONLY module that measures road safety (correlated to accident), validated by
experts of French Accident Administration (cf. paper at congress PRAC 2010 « évaluation du risque
de sortie de route pour l’aide à la conduite ou le diagnostic d’infrastructure », J Brunet, P Da Silva Dias, G. Yahiaoui, Session 1 « caractérisation du risque routier »).
To read the Newsletter : https://nexyad.net/Automotive-Transportation/?p=1667
Summary :
– French National Projects : Initiative PME (ADEME) – Safety/Risk of Accident
– Visibility Measurement : Release v2.0 of VisiNex Onboard
– Risk & Safety Estimation in Driving : SafetyNex Software Module
– Nexyad expertise about Autonomous vehicle : ERTS2 2016 – 8th European Congress
– Future Presence of Nexyad on Congresses & Exhibitions (ADAS, Autonomous Cars & Insurance)
– Nexyad New Customers : Focus on VeDeCoM
This project will help to us to accelerate applications of road safety/ risk of accident estimation
in embedded connected devices (based on the NEXYAD software SafetyNex) : new vehicles and also telematics devices (aftermarket applications for car industry, insurance companies, etc …)
Contact NEXYAD if you need to measure both safety and eco-driving onboard : (sales@nexyad.net)
Demo of SafetyNex : https://nexyad.net/Automotive-Transportation/?page_id=441
NB : SafetyNex is THE ONLY module that measures road safety (correlated to accident), validated by
experts of French Accident Administration (cf. paper at congress PRAC 2010 « évaluation du risque
de sortie de route pour l’aide à la conduite ou le diagnostic d’infrastructure », J Brunet, P Da Silva Dias, G. Yahiaoui, Session 1 « caractérisation du risque routier »).
VisiNex Onboard computes a distance of detection (in meters) and the VQS number which represents the easiness of objects detection and image understanding.
Calculation is processed using two multidimensional functions :
– the MTF function : Modulation Transfert Function that gives the contrast (contrats of Michelson) of the scene among frequency (also called the Spatial Frequency Response).
The MTF function is measured by VisiNex, with a very high signal/noise ratio.
– the CSF function : that gives sensitivity of human vision to contrast among frequency
(NB : this CSF function is also a function of background luminance, ambiant lightning, and may also be different for categories of persons : children, « regular » adult, old people, trained people such as military fighter pilots, etc …). The CSF function was measured by searchers thats validated it from the observation of thousands of people (a representative panel). The most famous version is the NGAN model. NEXYAD is involved for many years in military applications and could develop and validate a very accurate CSF that takes into account many parameters (such as ambiant light, background, …).
A mathematical comparison of those two functions lets VisiNex Onboard compute :
– a distance of detection, recognition, and identification, applying the Johnson criteria
– the VQS number which is the quantity of MTF that is actually helpfull for human vision.
So in the end, we can say that VisiNex Onboard outputs are the results on deterministic computing on contrast measurement.
References :
http://www.normankoren.com/Tutorials/MTF.html (MTF and CSF)
http://research.ijcaonline.org/volume40/number2/pxc3877165.pdf (CSF NGAN)
http://www.precision-vision.com/index.cfm/feature/12/d–contrast-sensitivity.cfm (Michelson contrats)
http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA456742 (Johnson criteria)
VisiNex also computes image distortion and can display a distortion map.
High level of distortion lead to images that are hard to understand and that is a cause of bad road safety (as a lack of visibility is).
From 27th to 29th january 2016 at Toulouse in France took place the 8th European Congress « Embedded Real Time Software and Systems », centre de Congrès Pierre Baudis.
Gérard Yahiaoui, Nexyad CEO participated to a roundtable about Highly Automated Driving conducted by journalist Laurent Meillaud and organized by Louis-Claude Vrignaud (Continental); with Paul Labrogere (IRT System X), Jean-François Simeon (Akka Research), Franck Foersterling (Continental Automotive), Ernesto Exposito (CNRS Laas), David Lopez (NXP) and Gilbert Gagnaire (EasyMile).
Connected Car Insurance Europe 2016
Tues 12th – Wed 13th April • Tower Bridge Hotel, London
Actualise Connected Data Sets to Create Innovative Motor Insurance Products
Nexyad will present SafetyNex at the conference and on his booth.
http://www.tu-auto.com/connectedcar-insurance-eu/
Active Safety Europe 2016: ADAS to Autonomous
May 17-18, 2016 • Munich
Enabling The Next Phase in Auto Safety Through The Development of ADAS
Gérard Yahiaoui, Nexyad CEO part of the 40 speakers.
http://www.tu-auto.com/activesafety-europe/
Autonomous Vehicle Test & Development Symposium 2016
May 31 to June 2 • Stuttgart
The world’s only conference dedicated to the testing and development procedures for autonomous vehicles and advanced driver assistance systems.
http://www.autonomousvehiclesymposium.com/
Connected Conference Paris
May 25-27, 2016 • Paris
Where Industry & Internet Meet
http://connectedconference.co/
CESA 4.0 – Automotive Electronics
16 & 17 November 2016 • Paris
Today’s automotive electronics & future trends and technologies.
Gérard Yahiaoui, Nexyad CEO part of Commitees
http://www.sia.fr/evenements/43-cesa-4-0-automotive-electronics?lng=fr
Save the dates, everybody is invited to come visiting NEXYAD…
The objective of VEDECOM is to become an institute for mutual and co-located research on electric, autonomous and connected vehicles, and on the mobility eco-system built on infrastructures and services addressing new usages of shared mobility and energy.
VeDeCoM has presented his Autonomous Vehicle at ITS Bordeaux in november 2015. Nexyad made a short video :
To know more : http://vedecom.fr/en/
The NEXYAD team is proud to announce that The VeDeCoM Institute bought user licences,
under RT-Maps Framework, of our Road Detection software module RoadNex and Obstacle Detection ObstaNex. |
Nexyad RoadNex v2.2 is the only Road Detection module software operable in framework RT-Maps by Intempora.
It is possible to connect multiple sensors included cameras on RT-Maps, collecting data in real time and time-stampted them. Later play and replay video with all your data.