Official codebase for Spotter: Let the Embodied Model Lead, and the VLM Reflect for It
Rethinking World-Action Model for Compositional and In-context Robotic Manipulation
Official code for ForesightSafety-VLA (IROS 2026): a unified diagnostic safety benchmark for vision-language-action models on RoboTwin
TronCamp ACT manipulation hackathon kit — LimX Tron2 dual-arm, RoboTwin, single-GPU ACT training, 4-task graded leaderboard (T4 stack_bowls_three).
Fast-VLA (FastVLA): You Only Think Once for Robot Manipulation
Official code repository for SkillNet
Give your agent a body: Embodify lets it control robots directly and become your embodied agent. Your best embodied agent is your favorite agent — with its memory, its tools and everything it knows about you. Works with Codex, Claude Code and any other agent that supports MCP and skills.
Evaluate JEV robot control in MetaWorld and RoboTwin: discrete actions, hierarchical control, privilege-level ablations, and replay analysis.
Compare GPT-6 Astra, GPT-4.1 mini, and Jev controlling an xArm7 in MuJoCo via direct Cartesian commands.
Exact full-corpus validator and evidence for the LeRobot RoboTwin dataset-to-Gym action contract.
LingBot-VLA 2.0 on RoboTwin: 50-task clean-only fine-tuning, resumable training, reproducible simulation evaluation, and RTC inference notes.
蚂蚁灵波具身大模型挑战赛（天池）初赛参赛记录：LingBot-VLA 2.0 × RoboTwin 50 任务 clean-only 后训练、评测管线与复现文档
RoboTwin / VLA rollout 记录汇总与 SHA-256 文件校验。
LingBot-VLA 2.0 x RoboTwin 2.0 LoRA fine-tuning: code, configs, eval results (62.8% clean / 37.0% randomized)
To associate your repository with the robotwin topic, visit your repo's landing page and select "manage topics."