Use when building, fine-tuning, deploying, or safety-wrapping Vision-Language-Action (VLA) models for robot manipulation — RT-2, OpenVLA, pi0, Octo, ACT, diffusion policies, ALOHA-style teleop data collection, action chunking, latency budgeting. Provides exact model specs, inference-rate math, worki
Use when building, fine-tuning, deploying, or safety-wrapping Vision-Language-Action (VLA) models for robot manipulation — RT-2, OpenVLA, pi0, Octo, ACT, diffusion policies, ALOHA-style teleop data collection, action chunking, latency budgeting. Provides exact model specs, inference-rate math, working OpenVLA/ACT/pi0 deployment code, dataset formats (RLDS/LeRobot), the failure modes that ruin runs