Designing for Assisted Driving
10 min read
Role
Master's Thesis Student
Responsibilities
User research, Interaction design, Prototype development
Managers
Duration
7 months
Overview
I had the opportunity to perform my Master's thesis project at VTI – The Swedish National Road and Transport Research Institute in Gothenburg, Sweden. The goal was to develop a dashboard interface that effectively communicated the operational limits of Adaptive Cruise Control (ACC) to drivers, enhancing safety and usability.
The work was presented at the SAFER Seminars and the Human Factors Collegium on 'Automation and Interaction Design' held in Gothenburg. A research abstract was also part of the proceedings of the 15th European Automotive Congress, held in Bucharest, Romania - November 2015.
Key Contributions
Driver Research
Performed comprehensive study to understand drivers' needs and preferences regarding the communication of technical failures in ADAS.
Improved Driver Awareness & Safety
Designed an intuitive HMI that clearly communicated ADAS system failures based on mental model metaphors, reducing confusion and improving driver response times.
Simulator Prototyping and Testing
Conducted simulator prototyping and usability testing with drivers, to ensure users can quickly interpret system feedback and react accordingly.
Insight towards HMI Design
Offered insights into effective visual communication for safety-critical systems for future automotive UX best practices.
Background
An ACC offers drivers with a more relaxed form of driving by partly taking care of longitudinal control of the vehicle once activated. However, the driver has to actively monitor the status of the traffic while using this system, and resume manual control when needed, i.e even when the system fails to respond under different circumstances.
The project was an attempt at designing user experiences for conveying critical system failures while using ACC and enabling a smooth transition from semi-autonomous to manual driving modes and vice versa.
Problem Statement
The limitations of Adaptive Cruise Control (ACC) systems have led to delayed driver responses, increasing the risk of collisions and failed manual takeovers when the system malfunctions. To enhance safety, there is a critical need for improved Human-Machine Interfaces (HMIs) that effectively communicate system failures and support seamless transitions between automated and manual driving modes.
The Solution
The solution was an instrument cluster concept with mode transitions integrated. The design used colour coding, earcons and informing texts in accordance with the concept of affordances and mental model methaphors to convey different system modes to the driver. Different colours for different modes such as color green for the manual driving mode, blue for semi-autonomous or ACC OK mode and red for the System failure or ACC failure mode were used.
The Process
The entire project was divided into 3 phases- Research, Design and Evaluation- Fairly standard phases for a design project. However, I adopted a modified simulator-based iterative design methodology, tailored for scenarios involving drivers as users.
Research Phase
The outcome of the research phase were the functional requirements and design guidelines for concept generation. The data picked up from the research phase were used in the Simulator-based design process (by Alm, Alfredsson & Ohlsson)- which began by creating paper sketches, moving onto building a desktop prototype and finally setting up the high fidelity simulator prototype using Processing for evaluations.
Design Phase - Simulator-based Iterative Design Methodology
The SBD process is presented below. The dotted lines represent iterative steps. For instance, once the pilot tests (see pilot tests in the flow diagram) were performed with some of the employees at VTI, I had to tweak the application codes to fit the scenarios.
Design reviews were a norm after each prototype development stage. Feedback were used in further iterations of the design before it was setup for the simulator study.
Evaluation Phase
The evaluation phase included an Icon intuitiveness Test, Heuristic evaluation and an in-depth interview, which were performed with 12 participants (6M +6F), aged between 25 and 51 years. All of whom had covered an annual driving distance between 15000 and 20000 kilometeres. In addition to the heuristic evaluations and in-depth interviews, notes were also taken during driving scenarios.
The introductory session, the driving scenarios and the in-depth interviews were all compiled as an audio recording for analysis. The data collection focussed on understanding the driver’s perspective of the warning function.
Concept Development
As a starting point, I was aiming to visualize basic layouts of the display. I used some keywords to describe different design directions and a moodboard for inspiration.
Snapshots from the concept development phase is presented below. The icon intuitiveness test needs a special mention. Generally, they are conducted in-person where a user is asked whether the graphic chosen for the icon represents the intended concept. Due to shortage of time and deadlines fast approaching, this usability test was conducted online as a survey within a span of few days. Surprisingly I received a fair amount of response- A total of 26 participants responded to the survey.
I chose the 'Space Agency' font family for the alerting and informing texts. The distribution of design elements was based on image schematic metaphors. That could easily help connect with the driver's mental model and reduce the learning curve for adapting to a new warning design.
All the relevant icons for "set distance", "set speed" and conveying semi-autonomy were re-designed and evaluated using an icon intuitiveness test.
Experimental Design
The final concept was tested in the driving simulator with 12 participants for different ACC system failure scenarios.
Participants were first introduced to the experiment and ACC functionality. They were provided with a printed paper prototype explaining the concept. A training scenario followed, where they were briefed on simulator limitations, such as the absence of gear shift, clutch, and secondary steering wheel functions. Finally, participants completed the experimental scenarios in a predetermined order.
The Driving Scenario
The driving environment resembled a clear summer day with a virtual representation of a Swedish rural highway with moderate traffic. The lead vehicle appeared on the road after an approximate drive of 150 seconds. ACC was activated when the driver accelerated to 70 km/h. Here, an audio notification (earcon) conveyed this activation. A distance of 3.0 seconds time interval to the lead vehicle was also pre-set. It was not possible to overtake the lead vehicle unless the driver applied brakes and switched to manual driving mode. The lead vehicle had a variable speed between 35 and 70 km/h during the car-following situation.
The lead vehicle braked several times during the car-following situation. The failure was injected during one of those times. For instance, in scenario A, the failure was injected when the lead vehicle braked for the third time, but injected the second time in scenario B. This was to avoid consistent events in two consecutive scenarios. The warnings (both audio and visual) were conveyed to the participant in scenarios B and C.
After the completion of all the scenarios, an in-depth interview was performed to understand the participant’s initial thoughts on what they experienced in the simulator.
Thoughts on driving with the warning information and without it, whether they preferred to add or remove any design elements from the interface and lastly thoughts on the earcons.
Results
Need for Continuous Info
Continuous information regarding automated functions is a necessity for the drivers.
Prioritise Auditory Modality
Drivers prefer warning information to be conveyed through the auditory modality in addition to the visual cues when engaged in a driving task even when automation is involved.
Avoid Visual Overload
Avoid displaying information related to primary driving tasks (longitudinal and lateral control of the vehicle) in the secondary display area during critical failure scenarios.
Takeaways
It's always wise to reflect upon your project jouneys. Here is what I learnt from it.
- Do not slack in the early stages of your project. This is a graph plotted using the approximate working hours I checked-in at the end of every project week. I thought it would be an interesting insight to see what project task I dedicated most of my time to. As the project progressed and the deadlines were approaching, I ended up spending more hours at the office, struggling to keep up with my project plan.
- More focus on the Project Variables. The sharp increase in average working hours during the project's later stages could be attributed to prototyping and testing. However, this might have been avoided with more thorough research on the simulator infrastructure earlier on.
- Documentation is as important as your results and analysis. I know many of you under value the importance of documentation throughout the product design process. The small note I used for the project proved invaluable during the analysis and reporting stages.
- Avoid egocentric intuition fallacy. Even though the process of sketching concepts and coming up with new ideas is fun, doing it alone can be hard when you have to make key decisions during the course of your project. There is always the risk of ‘egocentric intuition fallacy’ as suggested by Landauer (1997) which makes designers believe that their perception of a system is applicable to everyone else. Avoid this by getting as much feedback as possible throughout the project.
Interested in learning more about this project? here is the link to the project report. Feel free to reach out to me if you have any further questions about the project. Also check out this video I made showcasing some snippets from the study and what my participants thought of it.
Terragon Marketing Cloud
Helping over 7,000 African businesses retain customers at scale.
Nxtout.App
A social app to help students and professionals discover where they belong.
Multimodal Interaction Design
Exploring the potential benefits of multimodality for current and future in-vehicle experiences.
Sensibull Social
Crafting a Social Trading Platform