Benchmark RadarAI BENCHMARK PROFILE
RoboVista
RoboVista evaluates vision-language models on robot question answering with 474 visual question answering instances spanning 39 task types across agricultural, industrial, domestic, surgical, and autonomous driving domains.
- Released
- 2026-07-06
- Readiness
- Inspectable
- Primary field
- Robotics & Autonomous Systems
Why it matters
Robot applications require modular reasoning across diverse embodiments. This benchmark isolates decision components and shows correlation with real-world task execution, aiding VLM selection.
Motivation
Diverse applications for robotics, such as industry and agriculture, require robots to operate across various embodiments, changing visual conditions, and complex planning.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.