Popular scientific abstract
The advancement of technology, in various fields of research, is rapidly changing our environment; not the least, what we are used and familiar to. On of the most discussed topics today is the automotive vehicle industry, a field of research soon to revolutionize our whole traffic system. We are however facing an issue with respect to humans’ unpredictability. This poses a problem for the automated vehicle (AV) engineers because how do a machine account for what a pedestrian, driver or other vehicle is intending to do? For this reason, a lot of resources are put to study the interaction between human-machines. The current project aims to link the automated vehicle system to the humans’ natural cognitive framework in order to bridge the gap between them. By this attempt, risks of incidents and other uncertainties are expected to be reduced and thereby promote a safer traffic system. In phase two of the research project, we will account for individual differences and how they influence decision-making and evaluation of certain urban traffic state. By this, we expect to shed light into further directions of research and problems that need to be proceeded with caution. In sum, the results of my research will aid the engineers to produce appropriate models and designs of vehicles that take into account humans’ vulnerabilities and capabilities.
My affiliation
Supervisor: Professor Martin Baumann, martin.baumann@uni-ulm.de
Background
Despite the long way research have come with respect to AV, there still remains a barely touched area of importance: the human factors. As humans are unpredictable, it exponentially complicates matters for how an AV would interpret and account for it. Less complicated would be if all traffic were handled like robots and advanced machines in industrial warehouses. But this is not possible, namely because of our uncertainties. To account for this, there is an urgent need to understand the cognitive aspects of the driver when sharing the control of the vehicle. In some complex urban traffic situations, humans are still able to detect and account for risks better than the machines. In this early stage of this revolutionary era, a lot of focus must be put on the interchangeable sequence of taking and giving control.
My research therefore involves the interaction between the driver and the automated vehicle, hence the project title ”Internal Interface for Transparent and Agile Automation”. In other words, the project aims to develop cognitive models of dynamic task sharing between the driver and the AV. This will be done by accounting for individual differences (age, personality traits, cognitive load etc) as it is highly documented that these factors influence our driving behavior and risk assessment. I have the pleasant opportunity to work within SHAPE-IT at the University of Ulm, Germany. My academical background consists of a bachelor’s degree in cognitive neuroscience with the major topic in philosophy of mind, obtained from the University of Skövde, Sweden and a master’s degree in cognitive science with the major topic in human factors, obtained at the University of Linköping.
Aims and objectives
Given the complexities that arises when human interacts with AV, the aims at this stage of the overall project is to analyze the information needs of drivers. This is important as to ensure a transparent AV in dynamic urban traffic situations. This project also serves to get a deeper understanding of the degree of influence by the drivers’ characteristics (e.g. age, cognitive load, situation awareness) and to what degree would the interaction strategy vary due to these differences. Our objective therefore is to, based on analysis of collected data, recommend a human-machine-interaction (HMI) concept for transparent and dynamic task-sharing — and at the same time account for relevant individual differences.
The research project involves the interaction between the driver and the AV. As we still are in the initial step of having fully automated vehicles, current AVs and AVs in the near future will face many situations, especially in the urban traffic context, they are not fully able to handle without the support of the human driver. Probably, this required support will only consist in rare cases of taking back the complete control over the driving task. In many cases the automation simply will need the human driver to closely monitor its operation for a certain manoeuvre, to make a decision which manoeuvre to execute or to provide additional information to the automation. Therefore, what is required in these cases is the possibility of a flexible and dynamic task sharing between the human driver and the AV. This is only possible if the human drivers at each moment in time understand the tasks and responsibilities assigned to them, in which state the automation is and what tasks it can perform. The automation has to be transparent. Therefore, the research goals of this project are to understand the cognitive processes underlying the shifts in control between driver and automation and to develop human-machine interaction strategies that create a transparent automation and that enable dynamic task shifting. These will create a joint driver-vehicle system that is more efficient, safer and more comfortable than an automation mainly aimed at substituting the human driver in certain contexts given optimal conditions for the automation.
Results
This project was focused on internal interfaces for transparent automated vehicles in complex urban traffic scenarios. Mohammed was involved in the planning of an empirical study assessing drivers’ gaze behaviour and psychophysiological markers of stress and mental workload (Figalová et al., 2022). This study assessed cognitive and behavioural effects of predictability of control transitions during SAE L3 driving. The results of this driving simulator study suggest that an internal ambient interface increasing predictability of a takeover request resulted in improved performance without increased mental workload. Mohammed also started to work on a description of elementary cognitive processes involved in the perception of complexity in urban traffic. This preliminary concept and its discussion with ESR1 were one relevant source for the planning of the empirical study of ESR1 (Figalova et al., 2022). Additionally, some highly relevant results related to the topic of perceived complexity in general have been achieved in the work of Yuan-Cheng, with whom ESR6 exchanged ideas on complexity and its effects on perceived transparency.
Based on the results of Figalová et al. (2022), it is recommendable that the design of future automated vehicles includes ambient interfaces. Using the peripheral visual channel has proven to be efficient in continuously delivering additional information without increasing drivers’ mental workload. The peripheral channel used for ambient interface does not require focal vision, as opposed to traditional human-machine interfaces. Moreover, the work of Liu et al. (2022) recommends the design of transparent human-machine interfaces and a thorough, user-centred evaluation of driver-vehicle interaction strategies.