#RL (15)

The ultimate guide to multi-harness RL: train models inside Claude Code and Codex — Hugging Face's four-harness experiment with a 2.6B model

The same model behaves like two different models depending on the agent harness around it: GLM-5.2 scores 23% in one harness and 52% in another on SWE-bench Pro; in a controlled experiment swapping the harness moved scores 13 points while swapping the model moved only 2.5–5. A Hugging Face team (Adithya S Kolavi, Joel Niklaus, Lewis Tunstall, Leandro von Werra and colleagues, with Liquid AI) publishes an open fix: run agentic RL inside the real harnesses — Claude Code, Codex, OpenCode, Mini-SWE-Agent — with zero harness code changes. The stack: OpenEnv as the shared interface, a capture proxy that mints a session id per rollout and records exact token ids, per-token behavior logprobs and loss masks at the model endpoint (full-distribution sampling, top_p=1.0), Harbor for 40+ harness adapters and 26 sandbox backends, and TRL's Async GRPO consuming the masked training sequences. The numbers: LFM2.5-2.6B trained across four harnesses rises from 42.2% to 54.2% average pass@1 with gains under all four, and uses 31% fewer tool calls on tasks both it and the base model solved, while the OpenCode-only arm's gains stay mostly at home. The article also reports the SFT comparison (RL 54.6% vs SFT 47.5%), the full attribution of the earlier Qwen rise-and-decline (output-budget exhaustion, working without submitting, and a reward term that collapsed held-out accuracy from 0.740 to 0.178), and engineering boundaries like the proxy's concurrency ceiling.

BLOG

Reinforcement LearningRLAgent HarnessOpenEnvHarbor

0:45

Workshop: making real-world RL practical

@yigitkkorkmaz · 790 views · 2026-08-31

RLReinforcement LearningReal-World RL

Training a Humanoid Robot for Hard Work

Boston Dynamics shares how Atlas learns whole-body lifting via RL: from reference trajectories and reward design to GPU-scale simulation and real-hardware iteration, demonstrating fundamental building blocks of physical intelligence.

BLOG

AtlasRLProprioceptionBoston DynamicsReinforcement Learning

Atlas' Evolution From Research Robot to Industrial Humanoid

Boston Dynamics traces Atlas' decade-long evolution from hydraulic parkour research platform to electric industrial humanoid. 2026 production-ready Atlas unveiled with Hyundai, already deployed for part sequencing at the Hyundai metaplant.

BLOG

AtlasHyundaiRLBoston DynamicsHumanoid

3:13

REGRIND: Dexterous RL from Human Retargeting Initialization

@robotsdigest · 439 views · 2026-08-24

Dexterous ManipulationRLREGRIND

T-GMP: Terrain-conditioned Generative Motion Priors for Versatile and Natural Humanoid Locomotion

Achieving both anthropomorphic naturalness and robust terrain traversal remains a fundamental challenge in humanoid locomotion. Existing Reinforcement Learning (RL) approaches typically rely on fixed motion priors, limiting their adaptability to varying environments. We propose Terrain-conditioned Generative Motion Priors (T-GMP), a module that captures a terrain-conditioned latent motion manifold from a few expert state-terrain demonstrations using a Conditional Variational Autoencoder (CVAE). The learned priors enable smooth style transitions, facilitating a unified policy that adapts to terrain variations. We integrate T-GMP into an adversarial learning pipeline with our proposed Foothold Penalty, where a discriminator dynamically modulates naturalness constraints conditioned on local terrain features, guiding the generation of versatile and human-like motions. Experimental results demonstrate that our method outperforms existing baselines in traversal success rate and motion smoothness, while preserving biomimetically natural and physically coordinated motions.

Junhong Guo, Hao Hu, Chen Chen · 481 views · Jun 5, 2026

CVAERLLocomotionJun 5, 2026

1:54

Physics-Based Character Controllers with RL

@neural_avb · 340 views · 2026-08-09

RLCharacter Controlneural_avb

Light-Loco-Parkour: Versatile Perceptive Whole-Body Locomotion via Multi-Skill Distillation

Existing humanoid whole-body control systems still fall short of the way humans move through cluttered terrain: they either track expressive whole-body references without terrain generalization, or react to terrain online while leaving the arms, torso, and knees largely unused. We present Light-Loco-Parkour (LLP), an end-to-end perceptive whole-body locomotion system that closes this gap with a single deployable policy. Conditioned only on onboard depth and a velocity command, the policy decides when to walk, balance, climb, step down, or vault, with no reference input, skill label, hand-coded gate, or runtime motion graph. Compared with prior humanoid systems, LLP makes three contributions. First, it introduces a whole-body perceptive-control pipeline that extends an RL-trained, velocity-tracking locomotion policy with parkour skills learned from object-interacting motions, so the same policy tracks velocity in open terrain, executes whole-body traversal at obstacles, and resumes locomotion afterward. Second, it acquires terrain-conditioned skills from sparse seeds by expanding a single motion into dynamically feasible, terrain-paired references across obstacle geometry, rather than relying on a large motion corpus. Third, it learns autonomous skill transitions from reward, letting the policy decide when and which whole-body skill to invoke from depth and command alone, with no one-hot skill label, hand-coded state machine, or runtime motion generator. Simulation and real-world experiments show high success across both benchmarked terrains and unseen obstacle variations, and the same policy transfers zero-shot to indoor and outdoor hardware experiments. These results demonstrate autonomous perceptive whole-body locomotion on a humanoid in outdoor settings, using only onboard sensing and a single deployable policy.

Hongming Chen, Zhuoran Li, Hongxi Wang · 394 views · Aug 1, 2026

RLHumanoidWhole-bodyAug 1, 2026

2:33

PAC-MAN: Humanoid Dodges Balls for Whole-Body Safety

@yokoshaan · 365 views · 2026-08-02

PAC-MANHumanoidSafety

PAC-MAN: Perception-Aware CBF-RL for Whole-Body Safety in Humanoid Dodgeball

A perception-aware CBF-RL framework that couples control-barrier safety with onboard sensing for whole-body humanoid dodgeball. The deployed policy sees the ball only as segmentation-masked depth from a head-mounted camera, with CBF guidance representing clearance to every body link. Deployed zero-shot on Unitree G1, achieving 95% success rate on real-world throws.

Lizhi Yang, Junheng Li, Aaron D. Ames · 381 views · Jul 30, 2026

CBFRLUnitree G1Jul 30, 2026

0:31

Safe Execution of RL Policies Via Acceleration-Based CBF-QP Constraint Enforcement for Real-World Robotic Deployments

Reinforcement Learning (RL) has demonstrated remarkable capabilities for solving complex robotic control problems, but its lack of safety guarantees severely limits deployment on hardware. In particular, as legged robots and manipulators often operate near safety-critical boundaries, out-of-distribution states can lead to failure upon deployment. To address this, we introduce Acc-CBF-QP, an acceleration-based Quadratic Program (QP) safety filter using Control Barrier Functions (CBFs) that constrains any RL policy onto a safe set at runtime without modifying training. The method applies to unconstrained and Safe-RL policies, and enforces joint position, velocity, torque, and collision constraints within a unified optimization framework. A key contribution is the formulation of RL+QP tasks that regulate deviation from the RL command when constraints would otherwise be violated. We introduce a TorqueTask, minimizing torque deviation, and a Forward Dynamics Task, minimizing induced acceleration deviation, thus providing principled control over safety-performance trade-offs. Experiments on a 7-DoF Kinova Gen3 manipulator and a 19-DoF Unitree H1 humanoid, both in simulation and on hardware, highlight substantial reductions in constraint violations. On the real H1 hardware, a Safe-RL policy alone yielded 10.04 violations/s, which were reduced by 92% to 0.80 violations/s when augmented with Acc-CBF-QP. On the Kinova Gen3, Acc-CBF-QP fully eliminated violations. Nominal task performance of the RL objective is preserved in violation-free regimes. Under aggressive velocity commands on H1, Acc-CBF-QP improves execution by preventing constraint-induced shutdowns, yielding longer survival times. The full pipeline is open-source.

Bastien Muraccioli, Alice Cariou, Pierre-Alexandre Leziart · 353 views · Jul 16, 2026

RLSafetyCBF-QPJul 16, 2026

As an Amazon Associate, we earn from qualifying purchases.

#RL (15)

The ultimate guide to multi-harness RL: train models inside Claude Code and Codex — Hugging Face's four-harness experiment with a 2.6B model

Workshop: making real-world RL practical

Training a Humanoid Robot for Hard Work

Atlas' Evolution From Research Robot to Industrial Humanoid

REGRIND: Dexterous RL from Human Retargeting Initialization

T-GMP: Terrain-conditioned Generative Motion Priors for Versatile and Natural Humanoid Locomotion

Physics-Based Character Controllers with RL

Light-Loco-Parkour: Versatile Perceptive Whole-Body Locomotion via Multi-Skill Distillation

PAC-MAN: Humanoid Dodges Balls for Whole-Body Safety

PAC-MAN: Perception-Aware CBF-RL for Whole-Body Safety in Humanoid Dodgeball

WARL: Wrench-Augmented Reinforcement Learning (IROS 2026)

RoboParty humanoid disturbance-resilience framework

RL Framework Lets Robots Autonomously Choose Gait on Uneven Terrain

FlexionAI: RGB-Based RL Policy Training in Simulation

Safe Execution of RL Policies Via Acceleration-Based CBF-QP Constraint Enforcement for Real-World Robotic Deployments