Research Project for Autonomous Driving Transfer of Control between Driver and Autopilot Works

Mittels Kameras und Sensoren reagieren autonome Systeme abgestimmt auf das Fahrerverhalten.
Autonomous systems react to the driver's attention by means of cameras and sensors.

Autonomous driving can increase road safety - even if for the time being there will still be situations in which people can act with more overview than the machine. It is therefore crucial that control is reliably transferred from the vehicle to the driver.

How human drivers and digital pilots could work together in the future has been investigated over the last three years within the framework of the project » Personalized, adaptive cooperative systems for automated vehicles« (PAKoS) coordinated by KIT. One of the aims of the project was to identify critical driving situations and to transfer control to the driver in a forward-looking manner. In addition, an automated vehicle was developed that adapts to the needs of the driver.

In the future, when automated vehicles temporarily take over the entire vehicle guidance, it will become necessary to transfer vehicle control between the driver and the autonomous system - for example, before entering construction site areas with lanes and speed limits that are not clear for the system or before entering unmapped private premises. For a safe transfer of vehicle control while driving, the automated vehicle must recognize where the driver's current attention lies and can then adapt the driving style to the driver's activity.

The PAKoS system uses cameras in the interior and, if necessary, sensors on the steering wheel to identify the driver's condition and attention. It assesses the driver's current performance and his requirements for automated driving: Is the driver looking at the road or in a book? Is he talking into a mobile phone or is he digging in his pocket in the front passenger seat?

Such information and the necessary instructions for action are exchanged multimodally between the two cooperation partners, the driver and the vehicle, in order to support joint actions such as the transition between autonomous and manual driving. In addition, the driver can personalize the driving behavior with an app according to his needs. The prototype of the human-machine interface will now be presented in a demonstration vehicle.

The project in detail

The unique selling point of the PAKoS project is the development of a holistic and integrated concept for an individualised and personalised cooperation between driver and automated vehicle, which is necessary in order to be able to register future highly automated vehicles. Security and data protection as well as social aspects will also be considered in the project.

At KIT, three institutes contribute to PAKoS: The Institute for Control Systems (IRS) will be responsible for the technical implementation of the handover. The modeling of the haptic interaction between driver and vehicle and the resulting optimal support for humans are the central questions. The Institute of Anthropomatics and Robotics (IAR) deals with the video-based recognition of complex driver activities and interactions. The goal is the development of real-time methods under real-life conditions. The Institute for Technology Assessment and Systems Analysis (ITAS) coordinates and operates the technology assessment which closely accompanies the entire research and development process. In addition, generalized innovation-related recommendations are developed for decision-makers.

In addition to KIT and its partner FZI Forschungszentrum Informatik, the Technical University of Munich, the Fraunhofer Institute for Optronics, Systems Engineering and Image Evaluation, Bosch, BMW, the Spiegel-Institut Mannheim, Videmo Intelligente Videoanalyse, mVISE and Stadtmobil CarSharing are also involved. The project has a total budget of around 4.2 million euros and was funded by the BMBF with around 3.1 million euros.