Human-AI interaction
Master thesis at ABB
Exploring Interaction Concepts for Multi-Agent Orchestration in Industrial Automation Systems
2026 ·Mutli-Agent OrchestrationAI UXDesign Patterns for AI interfacesLanggraphExplainable AINDA
Thesis Work - Exploring Interaction Concepts for Multi-Agent Orchestration in Industrial Automation Systems
Future process control will increasingly rely on multi-agent systems that optimize objectives such as quality, energy, and throughput. Yet operators and engineers face challenges in understanding, supervising, and intervening when orchestration becomes complex or opaque. This thesis investigates how to design user-friendly, transparent, and adaptive interactions with multi-agent orchestration to reduce cognitive overload and enhance trust.
My role and responsibilities (according to the posting)
- Investigate how multi-agent orchestration can be performed, supervised, configured (e.g. adjustment based on human user’s objectives set-up) in process control contexts (e.g., view multiple agent objectives and trade-offs like balancing quality, energy, and throughput and intervene in orchestration, or trouble-shooting facilitation)
- Design and prototype interaction concepts (proof-of-concept) that illustrate how users (e.g. operators/system engineers) could have a user-friendly seamless process when working with multi agents where users may get cognitive overload
- Conduct concept validation sessions with domain experts or potential users to gather feedback on the concept
- Summarize findings into design implications for the broader multi-agent interaction framework
Lovable
- I inspired the UX team at ABB CRC to use Lovable and also held a presentation to the UX team on how I prototype with Lovable, tips & tricks, and how to get started.