Skip to main content
Back to challenges
Challenge

Build an Automated People-Mover Operations Agent with Google ADK

Managing high-density automated people-mover (APM) systems at Al-Maktoum International Airport requires continuous agentic monitoring. Build a multimodal monitoring system using Google ADK and Gemini 1.5 Pro to process live station camera frames and schedule passenger throughput.

Business OperationsHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Utilize Google Agent Development Kit (ADK) and Gemini multimodal vision capabilities to detect airport transit bottleneck locations and adjust train frequencies.

Delivery guide

How work is evaluated

Evaluation

Evaluates multimodal congestion detection accuracy and dynamic headway adjustment recommendations.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

headway_reduction_check

Ensure headway is reduced during CRITICAL density

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

density_detection_accuracy

Accuracy of platform crowd classification • target: 0.96 • range: 0.88-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Build multimodal observation tools using Google ADK and Gemini vision models

  • Detect platform passenger density from visual video feeds in real time

  • Automate dynamic dispatch of autonomous train cars to high-congestion terminals

  • Output operational telemetry logs compatible with airport control software

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic platform camera frames, passenger counts, and headway schedules for Al-Maktoum airport APM network.

Challenge setup

Evaluates multimodal congestion detection accuracy and dynamic headway adjustment recommendations.

Challenge input

JSON containing encoded frame and station details

Google ADK

Google standard agent kit optimized for Vertex AI and Gemini multimodal models

Evaluated output

JSON dispatch order

Evaluation checks
  • Ensure headway is reduced during CRITICAL density
  • Accuracy of platform crowd classification • target: 0.96 • range: 0.88-1
Configured evaluation evidence
  • Benchmark: MultimodalTransitBench
  • Density Detection Accuracy target: 0.96
  • 1 public reference case
Configured runtime
  • 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

Action Space
  • Google ADK · Required
  • Google · Optional
  • Google Gemini · Optional

Evaluation contract

  • headway_reduction_check · Weight 1
  • density_detection_accuracy · Weight 1

Metrics

  • density_detection_accuracy · Target 0.96

Benchmarks

  • MultimodalTransitBench
  • robotics

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing encoded frame and station details

    Output: JSON dispatch order

    Execution harness: Python

Evaluation tools

  • Google ADK · Agent Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Build an Automated People-Mover Operations Agent with Google ADK