Google ADK & Guardrails AI Model Hardware Standard Safe Device Router
Develop a hardware command router compliant with Model Hardware Standard (MHS) safety specifications using Google ADK and Guardrails AI. Implement physical payload validation rules that enforce strict actuator constraints, achieving zero unsafe state transitions across simulated robotic execution test suites.
What you are building
The core problem, expected build, and operating context for this challenge.
Construct an MHS-compliant physical hardware router using Google ADK and Guardrails AI to validate and control robotic actuator actions safely.
How work is evaluated
Tests safety boundary enforcing for physical hardware payload validation.
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.
out_of_bounds_rejection
Ensures all out-of-bounds hardware requests are explicitly rejected.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
guardrail_overhead_ms
Validation latency per command payload. • target: 5 • range: 0-50
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Implement Google Agent Development Kit (ADK) agent pipeline for MHS hardware command dispatching
Integrate Guardrails AI validators to enforce physical payload kinetic boundaries (velocity, torque, spatial bounds)
Build mock physical device driver interfaces conforming to Model Hardware Standard specifications
Achieve 100% detection and blocking of out-of-bounds mechanical control requests
Reference links and supporting material
Synthetic physical motor command dataset containing 500 boundary test scenarios across robotic arm, pan-tilt head, and conveyor systems.
Challenge setup
Tests safety boundary enforcing for physical hardware payload validation.
Challenge input
JSON object containing joint angles, motor torque values, and velocity vectors.
Google ADK
Structured tool call invocation and execution framework.
Guardrails AI
Provides schema validation for physical constraints.
Evaluated output
JSON safety response with status (APPROVED|REJECTED) and risk breakdown.
- Ensures all out-of-bounds hardware requests are explicitly rejected.
- Validation latency per command payload. • target: 5 • range: 0-50
- Benchmark: RoboBench Safe Action Standard
- Guardrail Overhead Ms target: 5
- 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
- Guardrails AI · Optional
- Guardrails AI · Optional
Evaluation contract
- out_of_bounds_rejection · Weight 1
- guardrail_overhead_ms · Weight 1
Metrics
- guardrail_overhead_ms · Target 5
Benchmarks
- RoboBench Safe Action Standard
- robotics
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON object containing joint angles, motor torque values, and velocity vectors.
Output: JSON safety response with status (APPROVED|REJECTED) and risk breakdown.
Execution harness: Python
Evaluation tools
- Google ADK · Agent Framework
- Guardrails AI · AI Engineering Tooling
Recipe freeze marker
No recipe freeze time is recorded.