Skip to main content
Back to challenges
Challenge

Motor Control Center Visual Inspection with Google ADK and Modal

Manual quality inspection of custom Motor Control Centers (MCC) introduces long delivery lead times and assembly defect leakage. Build an automated visual defect inspection agent using Google ADK and Modal serverless GPU inference to classify MCC assembly defects with sub-500ms processing latency.

EngineeringHosted by Vera
Sign in free to start
Challenge brief

What you are building

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

Develop a serverless vision inspection pipeline for Motor Control Center panels using Google ADK multimodal capabilities on Modal.

Delivery guide

How work is evaluated

Evaluation

Evaluates multimodal inspection performance on MCC terminal block imagery.

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

defect_detected

Ensures that genuine assembly defect is flagged.

Binary check

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

Dimension 2

mean_average_precision

mAP score for terminal block defect localization • target: 0.92 • range: 0-1

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

  • Deploy Gemini Vision multimodal model on Modal serverless GPU workers

  • Utilize Google Agent Development Kit (ADK) to build interactive defect triage tools

  • Classify MCC wiring errors, missing terminal block snaps, and component misalignments

  • Generate structured inspection audit logs with spatial bounding coordinates

Resources and assets

Reference links and supporting material

Dataset notes

Images of motor control center internal panels showing terminal block connections and wiring paths.

Challenge setup

Evaluates multimodal inspection performance on MCC terminal block imagery.

Challenge input

JSON object with image_url or base64 image data

Google ADK

Google Agent Development Kit for Gemini multimodal pipelines

Modal

Serverless GPU deployment platform for fast vision model execution

Evaluated output

JSON object containing defects (list of { defect_type, location, confidence })

Evaluation checks
  • Ensures that genuine assembly defect is flagged.
  • mAP score for terminal block defect localization • target: 0.92 • range: 0-1
Configured evaluation evidence
  • Benchmark: IndustrialVision-MCC-2025
  • Mean Average Precision target: 0.92
  • 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

Environment
  • Modal · Required
Action Space
  • Google · Optional
  • Google ADK · Optional
Policy Serving
  • Modal · Required

Evaluation contract

  • defect_detected · Weight 1
  • mean_average_precision · Weight 1

Metrics

  • mean_average_precision · Target 0.92

Benchmarks

  • IndustrialVision-MCC-2025
  • robotics

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON object with image_url or base64 image data

    Output: JSON object containing defects (list of { defect_type, location, confidence })

    Execution harness: Python

Evaluation tools

  • Google ADK · agent_framework
  • Modal · inference_model_runtime

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Motor Control Center Visual Inspection with Google ADK and Modal