Visual Forward Kinematics on Benchmark 10 visual problem instances 2 1.0 (test)
93Consistency ScoreRoboSolver (GPT-4o & Gemini 2.5 Pro Preview)
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| RoboSolver (GPT-4o & Gemini 2.5 Pro Preview)Version=Framework, Extractor VLM=Gemini 2.5 Pro Preview, Agent Model=GPT-4o2026.02 | 93 | 100 | 93 | 100 | 93 | 100 | — | 97 | |
| Gemini 2.5 Pro PreviewVersion=Raw2026.02 | 73 | — | — | — | — | — | — | — |