Skyline-Lifelines Hackathon 26'

Table of content

• About Us
• Problem Statement
• Solution Overview
• System Architecture
• Impact
• Evaluation Metrics
• Risks and Mitigation
• Use Cases
• Team
• References

🌇About Us

We are developing an AI-Driven Platform for Sustainable, Transparent Post-Disaster Reconstruction.

SKYLINE is an end-to-end software platform designed to transform post-disaster rubble into actionable rebuilding resources.

We integrate AI-powered damage assessment, ontology-based material reuse, and transparent procurement workflows and accelerate the recovery while embedding circular-economy principles into reconstruction.

🚧Problem Statement

Our problem: Large-scale disasters and conflicts generate massive amounts of damaged buildings and construction debris.

While modern computer vision can estimate damage and debris volume, current reconstruction workflows remain fragmented:

• Damage assessment is disconnected from rebuilding plans
• Debris is treated as waste rather than a recoverable resource
• Procurement and approvals are slow, opaque, and bureaucratically siloed
• Communities face prolonged displacement and rising reconstruction costs

We and SKYLINE propose an apporach to address this gap by converting damage data directly into decision-ready reconstruction actions

💡Solution Overview

SKYLINE is a modular, cross-platform system that links damage assessment → planning → material reuse → procurement → approvals in one coherent workflow.

Core Capabilities

AI Damage Assessment

• Computer vision models classify building damage severity
• Automatically estimate debris volumes with confidence scores

Reconstruction Planning

• Generates material quantities, cost estimates, timelines, and reuse potential
• Integrates historical building data and geospatial context

Ontology-Based Resource Engine

• Catalogs recoverable materials from debris
• Matches material condition, type, and location to reconstruction needs
• Turns rubble into a planning asset rather than waste

Transparent Procurement & Approvals

• Multi-criteria contractor bidding with auditable scoring
• Rule-based approval workflows with role-based access control

🧠System Architecture

SKYLINE is built as a modular, service-oriented platform:

AI Damage Detection Module

• CNN-based models for damage classification and debris estimation

Geospatial Layer

• Mapping APIs for geocoding, spatial optimization, and building data retrieval

Ontology Engine

• Semantic reasoning system for material matching and reuse optimization

Procurement Module

• Transparent, data-driven bidding and contractor selection

Approval Workflow Engine

• Licensing, audit logs, and compliance tracking

The MVP focuses on damage assessment, planning, and basic resource matching, with iterative expansion during pilot deployment

🌍Impact

Short-Term

• Reduce damage assessment time from days to hours
• Cut reconstruction planning and procurement from weeks to days
• Lower costs and accelerate reopening of critical infrastructure

Long-Term

• Institutionalize circular-economy practices in reconstruction
• Improve transparency, accountability, and community trust
• Enable scalable, resilient recovery across disaster contexts

📊Evaluation Metrics

SKYLINE is evaluated using: Time reduction in assessment, planning, and approvals

• Percentage of debris reused or recycled
• Cost savings per reconstruction project
• User adoption and satisfaction
• Transparency indicators (audit completeness, traceability)

⚠️Risks and Mitigation

Risk	Mitigation
Variable data quality	Confidence scoring + human-in-the-loop
Limited connectivity	Offline-capable design
Resistance to workflow change	Alignment with existing government & NGO processes

SKYLINE augments expert judgment, it does not replace it.

🧩Use Cases

Government reconstruction authorities Humanitarian and development organizations Post-conflict recovery agencies Urban resilience and sustainability programs

👥Team

• Jennie Nguyen (qdn@andrew.cmu.edu)
• Simon Malinka (smalinka@andrew.cmu.edu)
• Tram Tran (tntran@andrew.cmu.edu)
• Dilshodbek Khujaev (qzk@andrew.cmu.edu)

📄References

• Gupta, R., & Shah, M. (2021). RescueNet: Joint building segmentation and damage assessment from satellite imagery. In Proceedings of the 25th International Conference on Pattern Recognition (ICPR) (pp. 4405–4411). IEEE. https://doi.org/10.1109/ICPR48806.2021.9412295
• Trubina, N., Leindecker, G., Askar, R., Karanafti, A., Gómez-Gil, M., Blázquez, T., Güngör, B., & Bragança, L. (2024). Digital technologies and material passports for circularity in buildings: An in-depth analysis of current practices and emerging trends. In V. Ungureanu et al. (Eds.), CESARE 2024: Coordinating Engineering for Sustainability and Resilience (Lecture Notes in Civil Engineering, Vol. 489, pp. 690–699). Springer. https://doi.org/10.1007/978-3-031-57800-7_64
• United Nations Environment Programme. (2024). Sustainable debris management in Gaza. https://www.unep.org/topics/waste/sustainable-debris-management/sustainable-debris-managem ent-gaza