Raziskovalni projekti so (so)financirana s strani Javne agencije za raziskovalno dejavnost.
Raziskovalni projekti
Članica UL | UL Fakulteta za elektrotehniko |
Šifra | 1538 |
Naziv projekta | Nevrofiziološko in kognitivno profiliranje vozniških sposobnosti |
Obdobje | 01.05.2017-30.04.2020 |
Letni obseg | 1.59 FTE |
Finančni obseg | 400.000 EUR / ARRS (75%), Nervtech (13%), Zavarovalnica Triglav (12%) |
Vodja | Jaka Sodnik |
Veda | Tehniška veda |
Sodelujoče RO | Univerza v Mariboru, Filozofska fakulteta |
Vsebinski opis projekta | Glavni cilj predlaganega projekta je bil razvoj sistema za holistično evalvacijo vozniških spretnosti in predlog popravljenega vedenjskega modela voznika, ki temelji na izmerjenih dejanskih sposobnostih ter psihofizioloških, biometričnih in nevrofizioloških odzivih. Predlagani sistem je zgrajen na osnovi naprednega simulatorja vožnje na pomični platformi, ki zagotavlja doživeto in realistično izkušnjo vožnje v kontroliranem in varnem okolju. Slednji omogoča razvoj nabora raznolikih scenarijev vožnje na osnovi prometa, ki vključuje osebna vozila, tovornjake, kolesarje, pešce, itd. Ti scenariji združujejo normalne vozne pogoje z različnimi intenzitetami prometa, spreminjajočimi se cestnimi in vremenskimi pogoji ter tudi velik nabor nenavadnih, stresnih in celo nevarnih situacij, ki se jih v realnem življenju ne da replicirati. Pri tem merimo hitrosti, pospeške, obračanje volana, vijuganje znotraj voznega pasu, varnostne razdalje, reakcijske čase in druge lastnosti vožnje. Poleg podatkov zajetih in zbranih v simulatorju smo s pomočjo psihofizioloških, biometričnih in nevrofizioloških instrumentov opazovali tudi voznikov psihofizični odziv na stres in visoko kognitivno obremenitev.
Predlagali smo tudi lastno rešitev zalednega strežniškega sistema za shranjevanje vseh zajetih podatkov v poenoten podatkovni niz, ki je na voljo naprednim algoritmom strojnega učenja in modeliranja. Slednji omogoča označevanje in segmentacijo različnih visoko-nivojskih značilk z namenom predlaganja splošnega vedenjskega modela voznika, ki pojasni odzive voznika na različne situacije na cesti. Omogoča tudi identifikacijo različnih tipov in profilov voznikov ter tudi splošnih strategij v spreminjajočih se psihofizičnih razmerah. Identificirane so tudi posameznikove reakcije v različnih situacijah in vpliv stresa ter kognitivne obremenitve na vožnjo ter tudi neka splošna ocena vozniških sposobnosti. |
Sestava projektne skupine | Projektna skupina |
Faze projekta in njihova realizacija | Za povečavo klikni na sliko. |
Najpomembnejše objave |
|
Vse bibliografske reference | Povezava na SICRIS |
Fotografije | |
Research projects (co)funded by the Slovenian Research Agency.
Project
Member of University of Ljubljana | UL Faculty of Electrical Engineering |
Code | 1538 |
Project | Neurophysiological and cognitive profiling of driving skills |
Period | 01.05.2017-30.04.2020 |
Range of year | 1.59 FTE |
Financials | 400.000 EUR / ARRS (75%), Nervtech (13%), Zavarovalnica Triglav (12%) |
Head | Jaka Sodnik |
Research activity | Engineering sciences and technologies |
Research Organisation | Univerza v Mariboru, Filozofska fakulteta |
Abstract |
The main goal of the proposed project was to develop a system for holistic evaluation of driving skills and to propose an improved driver behavior model based on physical performance and psychophysiological, biometrical and neuropsychological responses. The proposed system was built around an advanced motion-based driving simulator providing immersive driving experience in a controlled and safe environment. The latter enables development of a variety of driving scenarios with traffic including cars, trucks, cyclists, pedestrians, etc. These scenarios include normal driving conditions with different varying traffic, road and weather conditions as well as a variety of unusual, stressful and even hazardous situations which cannot be duplicated in real life. Speeds, accelerations, steering control, lane deviations, safety distances, reaction times and other properties are measured and logged. In addition to data collected within the simulation a set of external psychological, biometrical and neurophysiological instrumentation was also used to assess driver’s psychophysical responses to stress and high cognitive workload.
A proprietary backend solution was proposed to store all captured data as a unique dataset available for advanced machine learning algorithms and modeling. The latter includes various high-level feature engineering and extraction techniques with the goal of proposing a general driver behavior model which will explain how drivers react in different situations on the road. It enables also identification of various driver types and profiles as well as common response strategies in varying psychophysical conditions. Individual’s reactions in different situations and impact of stress and cognitive workload on driving performance are also be identified and their overall assessment of driving skills is provided. |
Researchers | Link on SICRIS |
The phases of the project and their realization | Click on image to enlarge. |
Most relevant publications |
|
All bibliographic records | Link on SICRIS |