ESR8 – Human Factors in AI-based Automation Design

Popular scientific abstract

The world of transportation is undergoing a transformation thanks to advancements in automated technology, which promise to reduce road fatalities and improve mobility for both individuals and society. Automated Vehicles (AVs) hold the promise of reducing human-driver-caused crashes, potentially leading to fewer accidents, injuries, and fatalities. These vehicles also allow drivers to engage in other activities while traveling. This potential has spurred the automotive industry to rapidly develop AVs with varying levels of automation, from Advanced Driver Assistance Systems (ADAS) that support drivers to fully autonomous vehicles that manage all driving tasks under specific conditions.

However, despite their advantages, AVs present several challenges. Over-trust and over-reliance on these systems can lead to increased human workload and issues with driver engagement. These challenges are not limited to drivers but also affect other road users interacting with AVs. Unsafe and non-human-centered interactions can significantly diminish the benefits of AVs.

To address these issues and fully realize the benefits of AVs, it’s essential to incorporate human factors into their design. Human factors research focuses on understanding human capabilities and limitations and applying this knowledge to improve system performance, safety, and comfort.

Researchers advocate for integrating human factors knowledge early in the development process. Traditionally, this has been done through Requirements Engineering (RE), which involves eliciting, analyzing, documenting, and validating system requirements. However, the automotive industry’s shift towards agile development methodologies, which emphasize incremental and iterative development, poses new challenges. Agile methods often overlook human factors, focusing more on technical details.

Without a clear role for RE in agile development, integrating human factors knowledge as requirements becomes challenging. The lack of empirical research and clear guidelines further complicates this integration. Therefore, there is a need to develop new strategies to effectively integrate human factors in agile AV development.

Objectives

Given these challenges, my research aims to bridge the gap between human factors and agile development in the AV industry. This study is focused on developing practical methods and guidelines to ensure human factors are adequately considered throughout the AV development process.

The main objectives of this research are:

• Understand and describe how for example AI-based AV design can account for human factors in the agile way of working.
• Understand how requirements are communicated in industries having an agile workflow and how continuous experimentation is used, with focus on how and where HF requirements and knowledge are (and should be) considered.
• Identify and integrate disparate requirements from AV human factors researchers and designers of AVs, to improve road-user acceptance, AV transparency, and vehicle safety.

Who am I?

I am Amna Pir Muhammad, a Ph.D. student at the Chalmers | University of Gothenburg. Currently, I am working under the supervision of Dr. Eric Knauss and Dr. Jonas Bargman on a Marie Skłodowska-Curie Action Innovative Training Network project.

My research focuses on bringing Human Factors (HF) knowledge to AV developers. I believe it is vital to incorporate HF knowledge into AV development to make it more reliable and efficient. Before my Ph.D., I specialized in Software Engineering and worked as a lecturer at the Comsats University of Islamabad.

My passion lies in shaping the relationship between humans and AI-based autonomous systems, particularly in the fields of autonomous vehicle development, large-scale agile development, requirements engineering, and human factors.

Supervisors

Eric Knauss:                         eric.knauss@cse.gu.se

Jonas Bärgman:                  jonas.bargman@chalmers.se

Alessia Knauss:                  alessia.knauss@zenseact.com

Results

The following is an overview of Amna’s work in relation to the objectives and expected results.
We started with an exploration of challenges of incorporating explicit human factors into agile AV engineering (Muhammad et al., 2023a). From this exploratory study, a key learning was that strong communication channels will be needed. Ideally, agile AV engineering teams should be enabled to uncover relevant human factors knowledge. While this makes it necessary to have human factors expertise in the teams, teams should also be required to reach out to additional experts, when needed.
Continuous experimentation is a well-established practice among agile software teams. We investigated whether these experiments could support the generation of HF knowledge in agile teams (Muhammad 2023b). We discovered that goals of HF experiments and SW feature experiments may differ and that not all HF experiments lend themselves to the necessary automation for continuous experiments. We encourage future research in this area.
Traditionally, HF knowledge would be incorporated as requirements into the development activities. We found this to become significantly more difficult in agile development, since agile methods de-emphasize up-front analysis and comprehensive documentation (Muhammad et al., 2023a). Agile methods however allow for strong direct communication and iteration of work artifacts, which brings its own advantages for integrating HF research and AV development.
Given the difficulties and potential, we recommended the definition of a requirements strategy for agile development, clearly describing organizational aspects (roles and ownership), structural aspects (information model and traceability model), as well as workflow aspects (concrete activities scheduled in agile workflows) of managing requirements in agile development (Muhammad et al., 2022). We now followed up with specific considerations for HF-based requirements in such a requirements strategy (Muhammad et al, 2024a).
Regarding recommendations on what information should be communicated between AV human factors researchers and AV engineers in agile organizations to improve road-user acceptance, AV transparency, and vehicle safety: Our first recommendation is to consider the direction of this communication (Muhammad et al., 2023a). It can be a good idea if HF experts ask critical questions towards the Engineering teams, instead of providing the answers directly. Also, it is important to consider whether HF knowledge should be pushed into engineering or pulled. A key consideration for any company related to AV development is about where to place HF expertise. Based on a qualitative survey, we chart the landscape of strategies for HF placement (Muhammad et al., 2024b).
We have further evaluated common challenges and practices for communicating and incorporating HF knowledge into agile AV development (Muhammad et al., 2023c). These practices and associated challenges we then used to describe recommendations for HF-specific requirements strategies (Muhammad et al., 2024a). Here, we in particular sketch the solution space: how can engineering organizations set up their roles and ownership of work artefacts to support and optimize the communication of HF knowledge? How can they set up appropriate templates, tooling, and traceability? How can they manage HF-requirements in their agile workflows? For each of these three areas, we sketch the spectrum of solutions derived from interviews with experts at automotive companies. We anticipate that this will be critical input for any organization that aims to optimize communication between human factors researchers and AV engineers in their specific context.

Publications

  • Muhammad, A. P., Knauss, E., & Bärgman, J. (2023). Human factors in developing automated vehicles: a requirements engineering perspective. Journal of Systems and Software205, 111810.
  • Muhammad, A. P., Knauss, E., Bärgman, J., & Knauss, A. (2023, September). Managing Human Factors in Automated Vehicle Development: Towards Challenges and Practices. In 2023 IEEE 31st International Requirements Engineering Conference (RE) (pp. 347-352). IEEE.
  • Muhammad, A. P., Knauss, E., Bärgman, J., & Knauss, A. (2023). Continuous Experimentation and Human Factors: An Exploratory Study. In International Conference on Product-Focused Software Process Improvement (pp. 511-526). Cham: Springer Nature Switzerland.
  • Muhammad, A. P., Knauss, E., Batsaikhan, O., Haskouri, N. E., Lin, Y. C., & Knauss, A. (2022, November). Defining Requirements Strategies in Agile: A Design Science Research Study. In Product-Focused Software Process Improvement: 23rd International Conference, PROFES 2022, Jyväskylä, Finland, November 21–23, 2022, Proceedings (pp. 73-89). Cham: Springer International Publishing.
  • Muhammad, A. P. (2022). Managing Human Factors and Requirements in Agile Development of Automated Vehicles: An Exploration (Licentiate dissertation, Chalmers Tekniska Hogskola (Sweden)).
  • Heyn, H. M., Knauss, E., Muhammad, A. P., Eriksson, O., Linder, J., Subbiah, P., … & Tungal, S. (2021, May). Requirement engineering challenges for ai-intense systems development. In 2021 IEEE/ACM 1st Workshop on AI Engineering-Software Engineering for AI (WAIN) (pp. 89-96). IEEE.
  • Muhammad, A. P. (2021). Methods and Guidelines for Incorporating Human Factors Requirements in Automated Vehicles Development. In REFSQ Workshops.

SHAPE-IT Deliverables

  • de Winter, J., Berge, S. H., Tabone, W., Yang, Y., Muhammad, A. P., Jokhio, S., & Hagenzieker, M. (2023). Design strategies and prototype HMI designs for pedestrians, cyclists, and non-automated cars: Deliverable D2. 5 in the EC ITN project SHAPE-IT
  • Berger, B., Zhang, C., Muhammad, A. P., Knauss, E. (2023). The use of AI in AV human-factors research and human-factors requirements in AI-based AV design : Deliverable 2.4 in the EC ITN project SHAPE-IT
  • Merat, N., Lee, Y. M., Peng, C., Figalova, N., Mbelekani, N., Muhammad, A. P., … & Yang, X. (2023). Design guidelines for acceptable, transparent, and safe AVs in urban environments: Deliverable 2.6 in the EC ITN project SHAPE-IT.
  • Figalová, N.; Nasser, M.; Jokhio, S.; Mbelekani, N.Y.; Zang, C.; Yang, Y.; Peng, C.; Liu, Y.C.; Muhammed, A.P.; Tabone, W.; et al. Methodological Framework for Modelling and Empirical Approaches (Deliverable D1. 1 in the H2020 MSCA ITN Project SHAPE-IT). 2021. Available online: https://research.chalmers.se/en/publication/524589.
  • Merat, N., Yang, Y., Lee, Y. M., Hegna Berge, S., Figalova, N., Jokhio, S., Peng, C., Mbelekani, N., Nasser, M., Muhammed, A.P., Tabone, W., Yuan-Cheng, L. & Bärgman, J. (2021). An Overview of Interfaces for Automated Vehicles (inside/outside) (Deliverable D2.1 in the EC ITN project SHAPE-IT). SHAPE-IT Consortium. DOI: 10.17196/shape- it/2021/02/D2.1

Contact information

Amna Pir Muhammad
Department of Computer Science and Engineering 
Chalmers | University of Gothenburg
email: amnap@chalmers.se
LinkedIn, Google Scholar , Twitter

Archive

Introduction video from 2020

References

Muhammad, A. P., Knauss, A., Knauss, E., Bärgman, J., Heyn, H.-M. (2024b): Integrating Human Factors into Development of Automated Vehicles. (Submitted to a journal).

Muhammad, A. P., Knauss, A., Knauss, E., Bärgman, J. (2024a): Requirements Strategy for Managing Human Factors in Agile Automated Vehicle Development, In 2024 IEEE 32nd International Requirements Engineering Conference (RE) 

Muhammad, A. P., Knauss, E., & Bärgman, J. (2023a). Human factors in developing automated vehicles: a requirements engineering perspective. Journal of Systems and Software205, 111810.

Muhammad, A. P., Knauss, E., Bärgman, J., & Knauss, A. (2023b). Continuous Experimentation and Human Factors: An Exploratory Study. In International Conference on Product-Focused Software Process Improvement (pp. 511-526). Cham: Springer Nature Switzerland

Muhammad, A. P., Knauss, E., Bärgman, J., & Knauss, A. (2023c). Managing Human Factors in Automated Vehicle Development: Towards Challenges and Practices. In 2023 IEEE 31st International Requirements Engineering Conference (RE) (pp. 347-352). IEEE.

Muhammad, A. P., Knauss, E., Batsaikhan, O., Haskouri, N. E., Lin, Y. C., & Knauss, A. (2022). Defining Requirements Strategies in Agile: A Design Science Research Study. In Product-Focused Software Process Improvement: 23rd International Conference, PROFES 2022, Jyväskylä, Finland, November 21–23, 2022, Proceedings (pp. 73-89). Cham: Springer International Publishing