Build Multimodal Construction Permit & Zoning Inspector with Google ADK
Complex urban redevelopments—such as London's King's Cross—require navigating thousands of pages of site plans, building regulation documents, and heritage preservation blueprints. Developers will use Google Agent Development Kit (ADK) to build a multimodal construction inspection agent. The agent leverages Gemini's multimodal capabilities to analyze engineering blueprints, site photographs, and local zoning bylaws simultaneously. It identifies missing safety permits, flags non-compliant structural modifications on historic structures, and calculates construction schedule risk.
What you are building
The core problem, expected build, and operating context for this challenge.
Construct a multimodal permitting and zoning compliance inspection agent using Google ADK and Gemini.
How work is evaluated
Evaluates detection precision of zoning code non-compliance and missing safety permits from multimodal project packages.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
Compliance Determination Test
Ensures correct flagging of non-compliant height threshold
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Permit Identification Accuracy
Recall of required heritage permits • target: 0.95 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Utilize Google ADK to process multimodal inputs (site images + zoning text docs)
Implement document visual processing pipelines using Gemini 1.5 Pro multimodal tools
Detect missing architectural permits and safety code compliance violations in site plans
Generate automated compliance verification audit reports for local planning authorities
Reference links and supporting material
Multimodal dataset of 40 urban redevelopment site plans, permit applications, and zoning requirement docs.
Challenge setup
Evaluates detection precision of zoning code non-compliance and missing safety permits from multimodal project packages.
Challenge input
JSON with site_name, proposed_height_m, max_allowed_height_m, structural_changes_description, historic_designation
Google ADK
Required framework for Gemini agent development.
Evaluated output
JSON with compliance_passed (boolean), identified_deficiencies, required_permits
- Ensures correct flagging of non-compliant height threshold
- Recall of required heritage permits • target: 0.95 • range: 0-1
- Benchmark: UrbanPermitBench
- Permit Identification Accuracy target: 0.95
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
- Google · Required
- Google ADK · Optional
- Google Gemini · Optional
Evaluation contract
- Compliance Determination Test · Weight 1
- Permit Identification Accuracy · Weight 1
Metrics
- Permit Identification Accuracy · Target 0.95
Benchmarks
- UrbanPermitBench
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON with site_name, proposed_height_m, max_allowed_height_m, structural_changes_description, historic_designation
Output: JSON with compliance_passed (boolean), identified_deficiencies, required_permits
Execution harness: Python
Evaluation tools
- Google ADK · Agent Framework
Recipe freeze marker
No recipe freeze time is recorded.