gr00t

FlashRT is a high-performance realtime inference engine for small-batch, latency-sensitive AI workloads. The flagship integration is production VLA control for Pi0, Pi0.5, GROOT N1.6, and Pi0-FAST. Also support llm e.g, qwen3.6-27B

Code for "ACG: Action Coherence Guidance for Flow-based Vision-Language-Action Models" (ICRA 2026)

Inspired by Tony Stark's robotic assistant Dum-E, the mission of this project is to create an intelligent, voice & vision enabled AI agent with robotic arm(s) capable of real-time human interaction, physical operations, and orchestration of a wide range of tools and services.

Experimental: Robot Control for Strands Agents.

LoRA fine-tuning wrapper for NVIDIA GR00T N1.7 on a single 24 GB GPU

Monitoring & profiling toolkit for NVIDIA Jetson running Isaac-GR00T N1.5 TensorRT inference - Prometheus metrics, Grafana dashboards, per-engine latency profiling

Track B SONIC SIM curriculum research fork: bounded eval, variant checkpoint provenance, and SIM-M2 gates for GR00T-WholeBodyControl

Levarage Humanoid WBC for simulated boxing Matches

An NVIDIA Isaac GR00T N1.7 policy driving a simulated Franka Panda in MuJoCo: a closed perception-to-action loop from a language instruction and two camera views. Zero-shot, simulation only, with measured success rate and failure modes documented.

NVIDIA Isaac GR00T N1.6 - A Foundation Model for Generalist Robots.

Fine-tune NVIDIA Isaac GR00T N1.7 (3B VLA) on a single 24 GB RTX 4090 without root: patches, install script, MuJoCo evaluation, open datasets, 1800 evaluated attempts as JSON

Research design: agentic RL + curriculum learning for GR00T humanoid whole-body control, bridging NVIDIA TAO's Cosmos/LoRA/agentic post-training stack. Fork of tao-skills-bank + GR00T submodules.

Bimanual handover on a Unitree G1_29DoF: replacing Dex3 hands with Brainco Revo2, retargeting a GR00T N1.7 policy, and recovering the task with simulation fine-tuning (Isaac Lab)

Sim-to-real robot learning experiments with NVIDIA GR00T on SO-101, including remote inference on Brev, systematic evaluation, failure analysis, videos, and documentation of methods for reducing the sim-to-real gap.

SomaOS BrainAgent - a human-brain-inspired embodied AI framework: a full neurocognitive loop over 22 brain regions, with global-workspace arbitration, episodic-semantic memory, endogenous drives, and a perception-cognition-action loop. (Robonix community entry)

Closed-loop benchmark of 9 VLA policies (GR00T N1.7, pi0/openpi, LoRA vs full fine-tuning) on BEHAVIOR-1K. Imitation learning / behavioral cloning with diffusion action heads in PyTorch, plus a reproducible eval harness, unattended training automation, and raw metrics.

SomaOS Brain Next - a safety-first lightweight robot-brain runtime: zero runtime dependencies, dual-layer action gating, VLA hot-plug, deterministic execution, and traceable decisions. (Robonix community entry)

Sinan Robo · 司南·机脑 — auditable index of open / open-weight VLA robot foundation models: licence, weights, params, target embodiments, one evidence URL per field

TARX robotics beachhead: Unitree H2 Plus + NVIDIA Isaac GR00T integration, local OS layer, and contribution tracker.

Physical AI on the robot's own compute: GR00T through TensorRT on Jetson AGX Thor, Isaac Sim evaluation of a 6-DoF arm, ROS 2 on Orin NX, and a correction ledger kept on record. An EmbodiedEdge Labs project.