Periodic Reporting for period 3 - FITDRIVE (Monitoring devices for overall FITness of Drivers)
Reporting period: 2024-03-01 to 2025-02-28
The aim of determining fitness to drive is to achieve a balance between minimising any driving-related road safety risks for the individual and the community and maintaining the driver’s lifestyle and employment-related mobility independence. Driving a car is a complex and dynamic task and there is a wide range of conditions that temporarily affect the ability to drive safely like consuming substances or fatigue. Professional drivers are particularly affected by fatigue. The results of different surveys worldwide show that over 50% of long-haul drivers have at some time almost fallen asleep at the wheel.
The FitDrive project was conceived to identify and prevent driving stress states for professional drivers (and consequently driving fitness) with artificial intelligence and machine learning techniques able to build a “usual" driving profile of each driver after some thousands of kms driven. Once the "usual" driving profile of a specific driver has been defined, the AI system is able to detect “unusual” behaviour (outside the "usual" range of parameters) and associate them with the most probable causes, such as fatigue, or other cognitive disorders. The behavioural function consists of a set of indicators representing the range of usual driving behaviour for each specific driver, which will be defined by the self-learning system during a period of ordinary driving. Deviations are determined by the comparison between daily driving results and the personal “usual” profile.
The concept FitDrive system provides a continuous screening of the driver's psychophysical capabilities, alerting him or her to potential impairment on the way: in fact, the abnormal variations detected by the Artificial Intelligence can be associated with early situations of sickness that are not yet apparent to the subject but are about to manifest. The system will continuously learn and adapt itself to the driver: this means that the more a subject drives, the more the system adapts to him/her and the more it is able to make precise detection of anomalies, and the system will also be able to follow changes in driving habits. With only 30 days of unsupervised AI learning in the pilots, in vans and trucks, the FitDrive system detected driver states with 84% accuracy.
The FitDrive pilots positively tested the possibility for roadside patrol officers to interrogate the vehicle wirelessly and thus focus on those vehicles that have shown recent “unusual” behaviour, making inspections more efficient and reducing the time that vehicles remain stationary. A further reduction of the controls' time will be achieved through a new and faster drugs screening method.
The FitDrive pilot data will be made publicly available for the industry and the research community on Zenodo.
For fitness-to-drive and fatigue detection to be homogeneous across the different systems and to allow for authorities to intervene, it is necessary to create regulation that defines standard levels of fitness. Another requirement is updated regulation making more vehicle data mandatorily available over the OBD-II port, avoiding proprietary protocols hindering data access. FitDrive has put large effort in awareness raising concerning these matters, through various channels including three meetings at the European Parliament.
The FitDrive project was conceived to identify and prevent driving stress states for professional drivers (and consequently driving fitness) with artificial intelligence and machine learning techniques able to build a “usual" driving profile of each driver after some thousands of kms driven. Once the "usual" driving profile of a specific driver has been defined, the AI system is able to detect “unusual” behaviour (outside the "usual" range of parameters) and associate them with the most probable causes, such as fatigue, or other cognitive disorders. The behavioural function consists of a set of indicators representing the range of usual driving behaviour for each specific driver, which will be defined by the self-learning system during a period of ordinary driving. Deviations are determined by the comparison between daily driving results and the personal “usual” profile.
The concept FitDrive system provides a continuous screening of the driver's psychophysical capabilities, alerting him or her to potential impairment on the way: in fact, the abnormal variations detected by the Artificial Intelligence can be associated with early situations of sickness that are not yet apparent to the subject but are about to manifest. The system will continuously learn and adapt itself to the driver: this means that the more a subject drives, the more the system adapts to him/her and the more it is able to make precise detection of anomalies, and the system will also be able to follow changes in driving habits. With only 30 days of unsupervised AI learning in the pilots, in vans and trucks, the FitDrive system detected driver states with 84% accuracy.
The FitDrive pilots positively tested the possibility for roadside patrol officers to interrogate the vehicle wirelessly and thus focus on those vehicles that have shown recent “unusual” behaviour, making inspections more efficient and reducing the time that vehicles remain stationary. A further reduction of the controls' time will be achieved through a new and faster drugs screening method.
The FitDrive pilot data will be made publicly available for the industry and the research community on Zenodo.
For fitness-to-drive and fatigue detection to be homogeneous across the different systems and to allow for authorities to intervene, it is necessary to create regulation that defines standard levels of fitness. Another requirement is updated regulation making more vehicle data mandatorily available over the OBD-II port, avoiding proprietary protocols hindering data access. FitDrive has put large effort in awareness raising concerning these matters, through various channels including three meetings at the European Parliament.
The FitDrive concept was developed in 4 pilot cycles. The first two focused on determining features that are relevant for impairments, with a special focus on the onset of fatigue. This focus on early detection, vs. detection of already present fatigue, allows for timely warnings for the driver. Cycle 1 was performed in driving simulators and Cycle 2 on a closed test track.
For Cycles 3 and 4, correlations of the features were made with features that are easily measured, without invasive sensors, e.g. vehicle parameters. Individual AI models of usual driving styles were created during Cycle 3. These models were then used in Cycle 4 tests to detect unusual behaviour, subsequently related to possible causes of fatigue onset of other driving impairments. Multimodal unsupervised AI technology was developed for these individualised detections.
Prototypes were developed of a cloud-based vehicle monitoring dashboard was developed, a telematics device, a smart tachograph for communications with police patrols, and a portable drug screening device. Specific data-collection software for driving simulators was adapted and improved. Business sustainability plans were developed for these and other components.
FitDrive has advocated intensively for standardisation of fitness-to-drive measurements, with policy makers, e.g. through the new Driving Licence Directive. A request of amendment of the EU regulation that mandates vehicle data availability over OBD ports was initiated with support from the transport sector.
A proposal was made for including fatigue management in professional driver training; and a fatigue management methodology was developed including easy-to-implement tools for transport companies.
The four test cycles were successfully completed, despite a series of technical issues also due to the non -availability of data from several kinds of trucks, and the final tests demonstrated the validity and the potential of the FitDrive system.
During the project implementation it was clear that the project results, which are public to encourage OEMs to use them, may contribute to improve road safety, but need a common agreement to define the parameters of the different levels of measured fatigue and the thresholds beyond which: i) An alert is issued to the driver; ii) The driver is suggested to stop and check her/his status (threshold that could be used in legislation for police controls); iii) The SAE3 system is informed that the driver is not deemed capable of responding to an intervention request. The Consortium ran an extensive campaign targeted to both major stakeholders and EU institutions, including an online questionnaire, targeted newsletters, online workshops with a large participation and a final event at the European Parliament, which was the third one held in that location with the presence of MEPs of the Transport Committee.
For Cycles 3 and 4, correlations of the features were made with features that are easily measured, without invasive sensors, e.g. vehicle parameters. Individual AI models of usual driving styles were created during Cycle 3. These models were then used in Cycle 4 tests to detect unusual behaviour, subsequently related to possible causes of fatigue onset of other driving impairments. Multimodal unsupervised AI technology was developed for these individualised detections.
Prototypes were developed of a cloud-based vehicle monitoring dashboard was developed, a telematics device, a smart tachograph for communications with police patrols, and a portable drug screening device. Specific data-collection software for driving simulators was adapted and improved. Business sustainability plans were developed for these and other components.
FitDrive has advocated intensively for standardisation of fitness-to-drive measurements, with policy makers, e.g. through the new Driving Licence Directive. A request of amendment of the EU regulation that mandates vehicle data availability over OBD ports was initiated with support from the transport sector.
A proposal was made for including fatigue management in professional driver training; and a fatigue management methodology was developed including easy-to-implement tools for transport companies.
The four test cycles were successfully completed, despite a series of technical issues also due to the non -availability of data from several kinds of trucks, and the final tests demonstrated the validity and the potential of the FitDrive system.
During the project implementation it was clear that the project results, which are public to encourage OEMs to use them, may contribute to improve road safety, but need a common agreement to define the parameters of the different levels of measured fatigue and the thresholds beyond which: i) An alert is issued to the driver; ii) The driver is suggested to stop and check her/his status (threshold that could be used in legislation for police controls); iii) The SAE3 system is informed that the driver is not deemed capable of responding to an intervention request. The Consortium ran an extensive campaign targeted to both major stakeholders and EU institutions, including an online questionnaire, targeted newsletters, online workshops with a large participation and a final event at the European Parliament, which was the third one held in that location with the presence of MEPs of the Transport Committee.
The FitDrive concept was successfully used to detect deviations from a driver’s usual driving. These deviations can subsequently be classified, with a certain probability, to a cause of unusual driving like fatigue onset, drug use or neurogenerative disease. Unlike most current fatigue detection ADAS, FitDrive uses individualised AI monitoring, because here one size does not fit all: each driver behaves differently to upcoming fatigue or other impairments.
The driver state was communicated to police roadside control equipment using DSRC technology, implemented in the developed smart tachograph. This allows authorities to select vehicles increasing the effectivity of the controls. Additionally, a portable drug screening device was developed for use at police controls achieving detection of THC and methamphetamine in only 1.5 minute.
Impairment detection systems need (EU) standardisation so that all ADAS can apply the same standard, which in turn can be used for new legislation that allows authorities to perform controls on fitness to drive. Current fatigue detection systems implement proprietary evaluation rules of fitness to drive. FitDrive has advocated actively for standardisation, including three sessions at the European Parliament.
The driver state was communicated to police roadside control equipment using DSRC technology, implemented in the developed smart tachograph. This allows authorities to select vehicles increasing the effectivity of the controls. Additionally, a portable drug screening device was developed for use at police controls achieving detection of THC and methamphetamine in only 1.5 minute.
Impairment detection systems need (EU) standardisation so that all ADAS can apply the same standard, which in turn can be used for new legislation that allows authorities to perform controls on fitness to drive. Current fatigue detection systems implement proprietary evaluation rules of fitness to drive. FitDrive has advocated actively for standardisation, including three sessions at the European Parliament.