Benchmark RadarVEHBench
VEHBench is a diagnostic benchmark for LLM-assisted vibration energy harvester design, featuring 763 tasks across four design roles: specification triage, verifier-guided search, corrupted-state recovery, and policy-conditioned selection. It uses an analytical physical oracle for scoring.
- Released
- 2026-07-20
- Readiness
- Inspectable
- Primary field
- General AI
Why it matters
The benchmark addresses the need for stage-local evaluation of LLMs in coupled physical design workflows, revealing stage-dependent capabilities and response-control patterns. It provides practical value for selecting and routing models in engineering applications, highlighting that no single model dominates all stages.
Motivation
Battery-free Internet of Things (IoT) requires iterative design of vibration energy harvesters (VEHs) under coupled physical constraints, while LLMs are emerging as interface layers for engineering workflows.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.