Joost de Winter

Prof. dr. ir. Joost  de Winter

Cognitive Robotics Department

Faculty of Mechanical, Maritime and Materials Engineering

Delft University of Technology

Mekelweg 2, 2628 CD Delft, the Netherlands

E-mail: j.c.f.dewinter@tudelft.nl

Biographical Information

Birth: March 7, 1979; The Netherlands.

Educational Background

2009, PhD (cum laude), Faculty of Mechanical, Maritime and Materials Engineering

Delft University of Technology

Dissertation title: "Advancing simulation-based driver training"

2004, MSc, AeroSpace Engineering

Delft University of Technology

Research Field: ‘Cognitive Human-Robot Interaction’

My research focuses on Cognitive Human-Robot Interaction, specifically on developing touchless interactions between humans and machines. While my primary focus is on the application of these interactions in car driving, my research has broad implications in aviation and the medical domain as well. As a member of the Department of Cognitive Robotics at 3mE, the cognitive aspect of human-robot interactions is of paramount importance to my work.

With the rapid advancement of technology, mechanical devices are evolving into computerized systems that use sensors to gather data, make decisions, and execute actions. Automated driving is a prime example of these developments, and it has the potential to greatly improve road safety. Unfortunately, human error is still a significant cause of road accidents, and automated driving is not a complete solution. While a driver in an automated vehicle does not need to control the steering wheel and pedals, they still need to be attentive to the road and the subsystems, set automation modes, and regain control when the automated vehicle exceeds its operational envelope. Thus, the need for a touchless supervisory role remains essential. 

However, there is a downside to automation. Humans often struggle with the supervisory task required in touchless interactions. Studies have shown that humans lose situational awareness, become distracted, over-rely on automation, or become overwhelmed if the automation fails. Therefore, my primary research question is centered on creating safe and productive human-machine cooperation.

Adaptive Automation. One of the most exciting innovations in this field is automation that adapts to the human state. By automatically switching control between human and machine, a synthesis of human and machine capabilities is achieved. For instance, if a driver is distracted, as measured using eye-tracking, the car could take over control. Other examples include automation that verbalizes what the human has or has not seen or adapts to human stress levels as measured with physiological equipment. Combined with gesture-based control, human-machine interface cooperation becomes entirely touchless. The ultimate research goal is to create a symbiosis of human and machine cognition.

Operator Assessment and Feedback. In addition, my research focuses on operator assessment and feedback. By analyzing the data collected by contemporary cars, we aim to identify a driver's strengths and weaknesses and provide real-time tactile, auditory, and visual feedback. Our vision is to understand why some drivers make errors and others commit violations, which has broad theoretical and practical implications.

Multiple Humans and Machines. Moreover, multi-agent human-machine cooperation is also a crucial research area. As vehicle-to-vehicle communication and cloud-based data exchange become increasingly common, road users will exchange perception and knowledge, leading to distributed situation awareness. Examples of multi-agent interactions include an automated car that shares its intentions with pedestrians or warns of an oncoming road user. Additionally, we are exploring the use of augmented feedback for inter-road-user cooperation, which allows for customized feedback to road users.

In conclusion, the challenge for the upcoming years is to determine how human and automated agents can exchange information while being supported by real-time feedback. Our research combines classical engineering methods with insights from psychology and computer science. Furthermore, privacy concerns regarding information exchange are taken into consideration in our research.