SmartmindSmartmind

Research

Four themes, one city

The Urban Simulation Lab groups its ongoing research into four themes. 4 projects are funded by the university’s SRC.

Themes and projects

01

Urban Science

  • Urban population modellingSRC-funded
  • Level of vegetation land fragmentation in the Western ProvinceSRC-funded

02

Disaster Management

  • A framework linking land use, accessibility, density and surface runoff as a decision tool for plannersSRC-funded
  • An urban flood modelling app testing how building regulations cope with uncertain rainfallSRC-funded

03

Transport & Mobility Systems

  • Road crash magnitude versus road network configuration, Kegalle
  • Built environment versus travellers’ mode choice, Colombo

04

AI and machine learning

  • A deep-learning framework for road inventory systems
  • Deep learning for road inventories and 3D urban simulation

The journey of Smartmind

Research turned into chatbots: urban-planning Q&A in 2023, flood and landslide warnings in 2026.

Current milestone year: 01 / 06

  1. 2023

    First chatbot launched at EXMO 2023

    Launched at EXMO 2023, University of Moratuwa. Trained on real district development plans. Visitor feedback fed a published paper.

  2. 2023, after EXMO

    Chatbot 2.0

    Built on the Hugging Chat language model with web search, tuned response filters, a user-friendly interface and a “Pro Planning Mode”. A custom version was made for Port City Colombo.

  3. 29 December 2023

    Meeting with NBRO and the University of Manchester, UK

    Agreed a research project on an AI chatbot for landslide risk: early warnings, evacuation protocols, resilient-construction details and risk information. Supported by a Manchester grant.

  4. May 2025

    To be confirmed

    Best Scientific Paper, The Colombo Conference 2025 (ICSD-2025)

    A paper on a collaborative approach to disaster management and community awareness, featuring the AI chatbot built for NBRO.

  5. 20–22 May 2026

    Participatory flood and landslide chatbot

    Presented at the 1st International Online Conference on Urban Sciences. WhatsApp-based, multilingual, multi-agent design with language, live-weather API and scheduler agents. It gives residents preparedness guidance and lets planners interpret risk-sensitive zoning. Linked to SDG 11 and 13, and indirectly 9 and 17. Published 15 May 2026 on Sciforum.

  6. 22–24 September 2026

    To be confirmed

    Street Cloud at the 4th IOER Conference, Dresden

    Citizens share ideas alongside street imagery. Generative and vision AI produce redesign scenarios. Planners and people iterate together. Supported by the Alexander von Humboldt Foundation and DAAD, hosted by TU Dresden’s Chair of Mobility System Planning.

How the 2026 chatbot is built

A WhatsApp-based, multilingual, multi-agent design. It gives residents preparedness guidance and lets planners interpret risk-sensitive zoning.

Architecture of the participatory flood and landslide chatbotResidents and planners reach the chatbot through WhatsApp. A multi-agent design with a language agent, a live-weather API agent and a scheduler agent produces preparedness guidance for residents and risk-sensitive zoning insight for planners.WhatsAppchannelLanguage agentMultilingualmulti-agent designLive-weather API agentWeather dataScheduler agentTiming of messagesResidentsPreparedness guidancePlannersRisk-sensitive zoning
Source: this.md, Chatbot journey, 20–22 May 2026 row.
Institution-specific chatbots handle large workloads quickly and reduce dependence on individual officials.
The team invites collaboration to build tailored chatbots for individuals and institutions.

Wider topics in the department

The department’s ongoing research page names seven further topics.

  • 01Population distribution simulation
  • 02Building density and zoning versus runoff
  • 03Site-level land use and transport interaction
  • 04An AI-based landslide simulation model for Sri Lanka
  • 05Land fragmentation and urban sprawl
  • 06Citizen-science city mapping
  • 07A web-based decision system for Indo-Lankan cities

Department ongoing research page (opens in a new tab)

4ONSE

Weather stationTo be confirmed

A low-cost weather station built on open hardware, open software, open data and open standards. It measures eight environmental parameters.

  • Arduino (open hardware)
  • istSOS (open software)
  • CKAN (open data)
  • OGC standards

A joint project with the University of Applied Sciences and Arts of Southern Switzerland (SUPSI). The department lists it, and the team has yet to confirm that Smartmind may claim it.