2 PhD's on Human-Automated vehicles and 1 on human-drone interaction

In a larger Dutch National Research agenda (NWA) project Perceptive Acting under uncertainty: safety solutions for autonomous systems, led by CWI, we identified 3 different subprojects. Subproject 1 deals with Human brain and behavior (understanding how human brains deal with uncertainty in the real world. Subproject 2 is about Uncertainty and risk-aware autonomous agents (creasing a fundamental understanding of how autonomous agents can cope with uncertainty). Subproject 3 deals with human-agent interaction (understanding how autonomous agents can cope with human uncertainty). For subproject 3, we are looking for 3 PhD candidates.

Automated systems are becoming increasingly more versatile, and the technology of self-driving cars and autonomous drones is developing fast, with systems at various levels of automation entering society.

Although automated driving are often promoted as a solution for safety, comfort, efficiency and mobility, driving automation has now shown to bring challenges from the human factors perspective. The urban traffic is a complex environment. Human traffic participants (car drivers, pedestrians, cyclists) constantly deal with uncertainties - 'has that car seen me?' or 'will that car give me right of way?', 'what will the other car do', but interact with traffic not only in a cognitive manner but rather in a more natural skilled-based manner with implicit cues and communication. Automated vehicles still lack the capabilities to interpret (complex) human behavior correctly and adapt their behavior accordingly, let alone understand the mutual interaction with the vehicle influencing the behavior of the human traffic participants and vice versa. Subsequently, especially in the early days of adoption of this technology, it is important that humans - the users of this technology - can understand its capabilities and limitations not only in a cognitive manner, but rather in a more natural way and seamlessly and successfully interact with it.

In this context, trust is a critical issue, both from the people sitting inside these vehicles and people who come across these vehicles in the outer world. People have been known to over-trust by letting the system operate unmonitored when it isn't safe - leading to accidents. On the other side of the equation, there is also a lot of mistrust revolving driving automation, and people are afraid, hesitant, or scared of this new technology or interact with it. In addressing this, trust calibration can help, and people need to understand what the system perceives or does, and which actions it takes as a result without reading manuals, understanding sensors or following courses. Especially since there will be many different systems that will improve their behavior based on data and AI, knowledge-based knowledge does not offer a solution. Transparency of an intelligent system through intuitive and effective communication (Explainable AI - XAI) can be achieved through systematic research through design, ecological interfaces and natural implicit communication.    

This transparency is needed from the perspective of various roles in which humans will be interacting with the automated systems. A driver (passenger) of an automated car would potentially need to intuitively understand what the car is doing, and in the case of an unexpected behavior and uncertainty, this would still hold by increasing transparency.

From the perspective of other road users (pedestrians, cyclists, other drivers), the question arises how one can intuitively understand whether the automated vehicle is aware of the presence of the other traffic entities, and how it will interact with them in complex traffic negotiations with uncertainties. This challenge is further compounded when such interactions need to be facilitated with road users with disability or (visual) impairments.

Lastly, from the perspective of remote operators, the user needs to be able to control the vehicle safely with an understanding of the situation and context the vehicle is operating in, all from a distance. Therefore, adequate control interfaces are required, not just to operate the vehicle, but to understand the environment of the car, and also potentially communicate with the occupant(s).

We are looking for PhD candidates to fill 3 positions on the following broad research topics:

As a PhD-candidate, you will work on designing and validating natural interfaces and behavioral characteristics for automation systems (vehicles, drones) in order to reduce uncertainty and avoid unpredictable behavior for the human operators/drivers (drone, vehicle) who sometimes needs to take control, and for the bystander who are confronted with these autonomous systems (vehicle, drone).

Being a PhD requires you to be able to work in an interdisciplinary project and investigate how automated systems should interact with humans in handling complex and uncertain situations. Your tasks will be to use Ecological Interface Design principles to design interfaces to facilitate an intuitive and seamless cooperation between the operator and drone. After which you will evaluate and validate the design in simulator and real-world experiments. We are looking for a candidate with a strong background in traffic psychology, industrial design, preferably know about Artificial Intelligence and experimental studies and statistics.

For supervision of the first 2 positions, your supervision will take part at the Technical University of Eindhoven (TU/e), which will be the basis of your work environment.

For supervision of PhD 3, NLR is collaborating with the Technical University of Eindhoven (TU/e), at which you can work one day per week. See this link  for more information on this vacancy.