Thesis Topic - Refresh Button
AI-augmented public space that is sensing change and transforming itself when the world around it moves on.
Dr.-Ing. Martin Bielik · Chair of Informatics in Architecture and Urbanism · Bauhaus-Universität Weimar
The weather changes every day, foot traffic shifts every season, a new tram stop reshapes movement patterns overnight. The public space stays exactly the same. This thesis imagines a system that senses the drift and generates a design response before anyone has to ask.
Summary
A plaza is designed for today's traffic, today's demographics, today's climate. But cities change constantly — a new school opens, summers get hotter, a cycling culture emerges — and the original design becomes quietly mismatched with reality. No one notices until the gap is obvious, and by then it may have grown for decades.
This thesis builds a monitoring system for a public space that detects when surrounding conditions change enough to warrant a design intervention and then generates one using AI. The student's core task is not designing the space but designing the trigger framework: what types of change matter, how much change is enough, and how a detected mismatch translates into a design brief. A physical mockup demonstrates the full cycle from sensing change to generating a response, reviewed by urban design experts.
Description
1. Problem Statement
Urban spaces are designed for a snapshot in time. A plaza is shaped by today's traffic counts, today's demographics, today's climate. But cities change constantly — a new school opens, a neighborhood gentrifies, summers get hotter, a cycling culture emerges, a pandemic reshapes how people use public space. The original design becomes increasingly mismatched with reality, yet no one notices until the gap is obvious — and by then it may have grown for decades. Urban design treats space as a finished product. But the city it sits in is a living process.
Generative AI can produce anything from photorealistic images to architectural proposals. These models are not equally capable in every domain, but they are improving fast. This thesis speculates on a near future where they can generate plausible design interventions for a public space on demand. If generating a proposal is essentially free, the bottleneck shifts: the hard problem is no longer producing a design — it is knowing when a design is needed and what it should respond to. Should the system react when summer doubles foot traffic and people sit on the ground for lack of benches? When a new school changes pedestrian flows? When a heatwave makes an unshaded plaza unusable? The student's task is not to design a public space — it is to design the system that monitors one and decides when and why to trigger a design intervention.
2. Expected Outcome
A small-scale physical mockup of a public space paired with a monitoring system that detects changing conditions and triggers AI-generated design interventions. The student must define and deliver three things:
1. The application. A real public space where conditions change meaningfully over time — across seasons, through shifts in usage, or due to changes in the surrounding context. The student identifies the space, maps what changes (and how), and argues why it is a good test case for continuous design.
2. The trigger framework. This is the core design challenge of the thesis. The student must define what the system reacts to:
- What types of change matter? Short-term (weather, time of day, events), seasonal (summer crowds, winter emptiness), long-term (new infrastructure, demographic shifts, changing mobility patterns)?
- How is change detected? What data sources, sensors, or proxies make it observable?
- How much change is enough to trigger a design response? Not every fluctuation warrants a new proposal — the student must define thresholds and priorities.
- What information does the trigger pass to the AI? A detected mismatch needs to be translated into a design brief.
3. The prototype. A physical mockup of the chosen public space — small-scale, reconfigurable — connected to the monitoring system. When a trigger fires, the system generates a design intervention through AI and communicates what should change and why. The prototype demonstrates the full cycle: sense change → assess mismatch → generate intervention → show the recommendation. An expert review with 2–3 urban designers or planners evaluates whether the triggers make sense and whether the generated interventions are meaningful.
3. Presentation & Portfolio Value
Final presentation: A live demonstration of the physical mockup responding to simulated changes — triggers firing, the system generating interventions, the mockup showing what should change and why. Accompanied by the trigger framework and expert feedback.
Career signal: This thesis demonstrates the ability to: connect urban data to design decisions; think about space as a dynamic system rather than a static product; build physical prototypes that make abstract processes tangible; and design monitoring and recommendation systems for urban management — relevant for smart city platforms, urban data startups, municipal innovation, and computational design practices.
4. Submission
Always required:
- Portfolio PDF (2 pages, A4) — visual summary: the chosen public space, the trigger framework, examples of the system detecting change and generating interventions, and photos of the physical mockup in action.
- 30-second video reel — a shareable teaser showing the full cycle: a condition changes, the system detects it, a design intervention is generated, and the mockup communicates the response.
- 5-minute explainer video — walks through the problem, the trigger framework, the physical mockup, a demonstrated scenario, and expert feedback.
Working prototype:
The monitoring and trigger system connected to the AI generation pipeline, runnable by the reviewer. A reviewer should be able to trigger a simulated scenario (e.g., "summer foot traffic doubles") and see the system detect the mismatch, generate an intervention, and communicate it.
Specification document:
Describes the trigger framework: what types of change the system reacts to, how change is detected, what thresholds define "enough change," and how a detected mismatch is translated into a design brief for the AI. Also describes how interventions are generated and how they are communicated through the mockup. Not implementation documentation — the logic of when, why, and what the system responds to.
Validation report:
Documents the expert review: who reviewed (2–3 urban designers or planners), what simulated scenarios were demonstrated, and their assessment. Reports whether the triggers make sense, whether the generated interventions are meaningful, and where the system's judgement breaks down. Method, data, and findings.
Supporting materials:
- Physical mockup of the chosen public space — required, not optional. The mockup is central to demonstrating the trigger-to-intervention cycle.
- Video documentation of the mockup responding to at least 2–3 different simulated scenarios.
5. Requirements
Baseline: Experience with AI-assisted coding workflows; understanding of public space design and urban spatial quality; interest in physical prototyping and data-driven processes.
Acquirable during the thesis: Working with AI design generation tools; urban data sourcing (APIs, sensors, open data portals); physical mockup and prototyping techniques; sensor integration and trigger logic.
6. Solution Space (deliberately open)
The thesis is deliberately open in how monitoring, trigger logic, and the physical prototype take shape. The mockup could be a laser-cut model with projected overlays, an augmented tabletop, or a digitally augmented maquette. Triggers could be driven by real sensor data, simulated data streams, or manual scenario inputs. The AI generation could produce plan-view layouts, street-level visualizations, or abstract spatial diagrams. The student is encouraged to find the approach that best demonstrates the trigger-to-intervention cycle — and to propose alternatives not listed here.
7. Project Stages
- Select the space — Choose a real public space where conditions change meaningfully. Map what changes: usage patterns, seasonal dynamics, surrounding context. Argue why this space is a good test case.
- Define the triggers — What types of change should the system react to? Build the trigger framework: what data matters, how is change detected, how much change is enough, and what information does the trigger pass to the AI?
- Build the mockup — Design and build a small-scale physical mockup of the chosen space that can represent different configurations and interventions.
- Connect the system — Link the monitoring logic and trigger framework to the AI generation pipeline. When a trigger fires, the system generates an intervention and communicates it through the mockup.
- Test & review — Run the system on simulated scenarios (e.g., "summer foot traffic doubles," "a new tram stop opens nearby"). Review with 2–3 domain experts: do the triggers make sense? Are the interventions meaningful?
- Documentation & presentation — Write the thesis. Prepare the live demonstration for the final review.
8. Keywords
continuous urban design, design triggers, public space monitoring, design mismatch, generative design, responsive urbanism, physical prototyping, smart city
Before You Apply
Read the supervision terms and conditions — how I work, the timeline and deadlines, the consultation structure, and what is due when.
Dr.-Ing. Martin Bielik · InfAU · Bauhaus-Universität Weimar