A space dedicated to experimenting with innovative solutions based on AI algorithms and blockchain techniques to enable new models of mobility and smart cities.
Harnessing the potential of high-fidelity simulations, Reply delves into the future of autonomous driving solutions developing an innovative AI-enhanced and sensor-less approach for identifying impaired driving conditions.
#Impaired driving detection
#Driver state monitoring
#Artificial Intelligence on the edge
#Traffic scenario-based testing
In order to reduce the number of road accidents caused by human factors, Driver State Monitoring (DSM) solutions have been developed. These systems are capable of detecting instances of impaired driving, such as drowsiness, distraction, driving under the influence, sudden illnesses, and aggressive driving. These solutions are also required by the European Regulation which, starting from 2026, it will be mandatory to equip all new vehicle types with the Advanced Driver Distraction Warning System (ADDWS).
Reply has developed a driver state monitoring system based on the analysis of behavioural patterns (without the need for biometric data), and the creation of a Digital Twin of the driver. This model of nominal behaviour is adapted to each specific driver's driving style by collecting telemetric data on an AWS platform and analysing the historical usage on the road.
To validate this algorithm, a testing framework has been developed, which includes an in-cabin monitoring system capable of analysing drivers' biometric data and validating instances of inappropriate driving activities.
Detects impaired driving without the necessity to tap into biometric information.
Create a Digital Twin tailored to each verified driver, reflecting their unique driving patterns.
Engineered to operate directly on the vehicle's ECU, minimising lag and bypassing network limitations.
Streamlines the data tagging and system verification process by using real-time insights gathered by an in-cabin camera monitoring setup.
Anchored on a unified vehicle connectivity framework, it's adept at transmitting telemetry insights to distant databases and facilitating over-the-air model updates.
Utilises computer vision techniques to derive biometric insights, ensuring precise assessment of driver distraction levels.
A synergistic, scalable co-simulation environment combined with a realistic vehicle model is the basis for a versatile tool that can be used in development and validation, especially for assessing traffic scenarios.
Generation of high-fidelity vehicle and traffic data from a simulation for automotive system development and validation.
As autonomous driving technologies continue to break barriers, the synergy between real-world and simulated data stands as a testament to the industry's commitment to innovation, safety, and progress. Specifically, Reply's solution embodies this forward-thinking approach, harnessing the power of high-fidelity simulations for advanced and personalised driver state monitoring.
Unlike real-world data acquisition which involves extensive sensorisation and experimentation, simulated environments can produce massive amounts of data in a fraction of the time and cost.
Simulated driving scenarios mitigate the risks tied to real-world anomalies, ensuring driver safety isn't compromised during data collection.
With simulations, one can reproduce diverse driving scenarios, ensuring a holistic dataset that mirrors various real-world conditions.
From capturing subjective behaviour with a real driver in the loop, to interactions with smart road infrastructure, simulations offer a detailed and realistic portrayal of on-road conditions.
The certification processes of these simulations are already aligned with EU GSR legislation, paving the way for more streamlined, legal, and standardised data collection processes.
The autonomous driving technology and high-fidelity simulation are tangible realities at Area42 labs: the Reply’s development centre is dedicated to experimenting with and transforming creative ideas into reality and it is where real prototypes can be felt, tested, and adapted to any business context.
This is the future of autonomous driving, and it's being shaped right now at Area42.