embodied-agent-arena/embodied-agent-arena
Seven frontier VLM agents, five robotic capabilities, 1,000 cases. Compare geometric and spatial reasoning, affordance grounding, task planning, and manipulation with source-native evaluation.
GitHub repository with 7 stars and 0 forks.
Language: Python
Topics: benchmark, embodied-ai, robot-manipulation, robotics, spatial-reasoning, vision-language-models