Gemini 4 Argon vs Gemini Robotics-ER 2 in Isaac Sim multi-robot assembly
In a three-arm, six-layer tower benchmark, Robotics-ER 2 finished 6/6 layers in 30.63s while Gemini 4 Argon stalled at 5/6 after about 16s, exposing workspace-geometry gating issues.
Multi-robot TAMP benchmark in Isaac Sim + ROS 2: Gemini Robotics-ER 2 vs. Gemini 4 Argon on 3-arm, 6-layer tower assembly. ☝️Robotics-ER 2: 6/6 layers in 30.63s (13.8s plan, 16.8s exec, 1,735 tokens). Respected shared-zone limits. ✌️4 Argon: Stacked 5/6 in ~16s, then stalled; skipped FR3_1 go_home retraction and dispatched FR3_3 to pick Block5 off FR3_1's table (out of kinematic reach). Proposing a fast parallel DAG is easy; gating for workspace geometry is the real bottleneck. ✍️Write-up: https://mincasurong.github.io/blog/ 💻Repo: https://github.com/mincasurong/IsaacSim_GeminiRobotics