The SituWare project set out to create a comprehensive software and hardware platform, SituSYS, that can reliably detect, interpret and incorporate a driver’s situational awareness into vehicle control. The goal was to improve hand‑over scenarios in highly automated driving by shortening transition times, reducing unnecessary warnings and thereby increasing driver acceptance and trust. A key technical challenge was to design adaptive interaction strategies that do not interfere with the driver’s chosen actions and do not distract the driver during the hand‑over.
AVL’s contribution focused on sensor selection, driver‑state algorithms, a hardware‑in‑the‑loop (HIL) and virtual‑reality (VR) testbed, a vehicle demonstrator and the integration of all components. The sensor suite is entirely non‑invasive. A primary camera monitors head pose, eye‑opening and gaze direction, while a secondary side‑view camera detects secondary tasks such as radio or phone use. Additional sensors improve robustness through sensor fusion: a seat‑position sensor indicates whether the driver is turning to reach for a child, steering‑wheel touch sensors confirm hand presence, and a heart‑rate and breathing sensor from a fitness‑tracker provides fatigue and distraction indicators. Hand‑on detection on the steering wheel is achieved either by a capacitive sensor in the test vehicle or by detecting steering‑wheel rotation in the simulator. The seat‑position sensor was supplied by a partner that had previously developed it for another project. No electroencephalography (EEG) system was adopted because none on the market met the required driver‑state detection performance.
The software architecture was built on the Robot Operating System (ROS) middleware, chosen for its ease of component exchange among partners. ROS enabled the seamless integration of AVL’s own modules with those from the other consortium members. A VR system was constructed to validate the driver‑state detection and adaptive interaction logic. The VR environment could simulate an “autobahn with construction site” scenario, varying weather conditions and traffic density. Attention detection was performed by analysing gaze direction, and the human‑machine interface (HMI) displayed real‑time attention and fatigue levels to the driver. The VR system also allowed the evaluation of the hand‑over strategy under controlled yet realistic conditions.
The project was organised into six work packages. Work package 1 defined scenarios, use‑cases and requirements. Work package 2 (led by AVL) handled sensor selection and driver‑state detection. Work package 3 (led by DLR) focused on modelling and interpretation of situational awareness. Work package 4 (led by the University of Ulm) developed adaptive interaction strategies. Work package 5 integrated all components into the SituSYS platform and evaluated it in both simulation and a real vehicle demonstrator. Work package 6 (not detailed in the excerpt) presumably dealt with system validation and dissemination.
SituWare was funded by the German Federal Ministry of Economics and Climate Protection as a national collaborative research project. The consortium comprised five partners: AVL Software & Functions GmbH, Deutsches Luft- und Raumfahrtzentrum e.V. (DLR), Human Factors Consult GmbH (HFC), Humatects GmbH (HMT) and the University of Ulm. AVL provided administrative leadership of the consortium, while DLR supplied technical leadership. The project ran for three years, with a two‑month cost‑neutral extension at the end.
In summary, SituWare delivered a fully integrated driver‑state monitoring and adaptive hand‑over system that combines multiple non‑invasive sensors, ROS‑based software architecture, a VR validation environment and a real vehicle demonstrator. The system’s design and implementation lay the groundwork for safer and more acceptable transitions between automated and manual driving, addressing a critical barrier to the widespread adoption of high‑automation vehicles.