BACK TO HOME

SPATIAL PROMPTING TERMINAL

LOCATION-AWARE AI ASSISTANCE

PHD RESEARCH

ONGOING

UdK Berlin

A person moves the Spatial Prompting Terminal through a shared learning and workshop environment.

HOW CAN AI BECOME SITUATED IN PHYSICAL SPACE?

The Spatial Prompting Terminal is an experimental research prototype developed as part of Johannes Pointner's PhD. It investigates location-aware AI assistance in physical learning and work environments. At its core is the question of how a large language model can become a spatially situated interface that not only processes linguistic input but also incorporates the concrete environment into its responses. The system uses UWB-based indoor positioning to track the mobile artefact in real time and link its location to defined spatial zones, work areas and object contexts. Users interact with the system through spoken or text-based requests and receive information, orientation or suggestions for action tailored to their current position and the spatial resources available there.

A person uses the mobile Spatial Prompting Terminal next to tools and fabrication equipment in a workshop environment.
The interface provides situated assistance in relation to its current location.

Situated AI Assistance

The prototype moves generative AI from the isolated screen interface into physical and situated interaction. Knowledge is not retrieved solely through abstract prompts, but emerges from the combination of language, place, proximity and context of action. The Spatial Prompting Terminal investigates how spatial position itself can become part of a prompt and what new forms of support emerge when AI can refer to local tools, infrastructures and situations.

Research Through Design

As a research-through-design prototype, the project makes questions of contextual understanding, intelligibility and the limits of spatially situated AI tangible and open to investigation. It explores what aspects of a physical situation can be represented computationally, how spatial context influences AI-generated responses, and how users can understand and control the relationship between location, retrieved knowledge and suggested actions.

A plan view visualises the terminal's position, surrounding furniture, spatial zones and relationships to nearby objects.
Spatial position is linked to zones, objects and local resources.

Role and Contribution

Johannes Pointner initiated and develops the Spatial Prompting Terminal as part of his doctoral research. He is responsible for the research concept, spatial and interaction design, physical prototype, UWB-based positioning, context model and system development.