Papers

QwenGyre: An Elastic Reinforcement Learning Framework for Training xLong-Horizon Agents

Large language model (LLM) agents increasingly undertake extreme-long (xLong) horizon tasks, where a single execution can span hours, hundreds of model-environment interactions, and nearly 1M tokens per rollout. Applying online reinforcement learning (RL) to such executions poses two fundamental challenges: (1) severe execution variance and prolonged rollout delays cause massive GPU idling; and (2) complex non-linear branching generates massive trajectory redundancy, crippling training efficiency. To address these, we present QwenGyre, an end-to-end framework for xLong-horizon online RL. QwenGyre elastically reallocates GPUs between rollout and training without interrupting live executions, while its trajectory processor reconstructs branching histories, scores partial progress, and deduplicates redundant paths to bound training costs. Scaled to our flagship model, Qwen 3.8 2.4T, with 700K tokens per rollout, QwenGyre yields a 6.0% absolute gain on NL2RepoBench (52.5% to 58.5%) in 48 steps. Across our evaluations on diverse domains of training datasets, QwenGyre delivers up to 1.85x and 1.78x speedups over Colocate and Async, respectively.

Reinforcement LearningLLM Agent

Weiqi Wang, Yuxin Zhou, Mouxiang Chen · 392 views · Sep 27, 2026

RSI-Anything: Unlocking Recursive Self-Improvement with World Models

RSI-Anything turns predictive world models into the steering signal of a recursive self-improvement loop built on JEPA-Anything. Predictions decide which candidate actions earn scarce external feedback; that feedback then drives policy learning or workflow revision and, when reliability checks show the predictor has drifted under the changed policy, conditional world-model adaptation, while component and compatibility gates decide which policy-model pair is retained to shape the next round. With the coder held fixed, hidden-test success on 200 CodeContests tasks rises from 26.3% under a capacity-matched static verifier to 29.2% with a fixed world model and 31.2% after adaptation, at comparable online test usage. GUI evaluation reaches 99% on MiniWoB++ click-checkboxes while cutting mean actions per episode by 39.4%; five successive laboratory-planning policy versions more than double the LabHorizon F200 Level-2 score from 21.80% to 46.01%; and in a 40-experiment Suzuki-Miyaura flow screen RSI-Anything reaches 96.15% mean best observed yield against 82.16% for Bayesian optimization. An end-to-end immune-science application carries the same retained-workflow-plus-predictor contract into prospective therapeutic hypothesis generation.

Recursive self-improvementWorld Models

Taoyong Cui, Pinxian Zeng, Wanghan Xu · 539 views · Oct 9, 2026

IdeaScientist: Orchestrating Agents for Grounded Scientific Ideation

Most autoresearch systems couple idea generation with implementation, experiments, and writing, so it is unclear whether success comes from the idea or from downstream execution. IdeaScientist isolates research ideation as a standalone task: given a problem definition and a challenge that current solutions fail to meet, produce a grounded research proposal. An orchestrator coordinates three sub-agents - a gap finder that locates the shared methodological limitation in same-problem prior work, an innovator that retrieves mechanisms which solved an analogous challenge in other fields and adapts them, and a report writer that develops the surviving intuition into a complete proposal. Each role is trained separately with asynchronous GRPO and reference-grounded rubric rewards. The Svalbard Idea Vault decomposes 4.7M arXiv papers into 2.77M (problem, challenge, gap, intuition, proposal) tuples for retrieval, training, and temporally controlled evaluation with a January 1, 2026 cutoff and 277 held-out post-cutoff methodology papers. On Qwen3.6-27B, IdeaScientist reaches 74.6% overall, 14.0% above the strongest open-source autoresearch baseline, driven mainly by novelty, and beats Claude Code SDK with Claude-4.8-Opus by 5.9% and Codex SDK with GPT-5.4 by 5.1% on the same open backbone.

AI ScientistLLM Agent

Jiarui Liu, Renjie Tao, Yiwei Liao · 647 views · Oct 2, 2026

Exposing the Long-tail in Embodied Urban Navigation via Scalable Learning from In-the-Wild Videos

Learning embodied urban navigation policies from real-world data is constrained by the cost of task-specific data collection and the limited coverage of rare yet safety-critical scenarios. To address these challenges, we present a scalable framework for learning point-goal urban navigation from web-scale in-the-wild egocentric videos while systematically exposing its long tail. The framework automatically annotates uncurated web videos with metric trajectories and structured navigation semantics, which are then used to train a vision-language-action policy for interpretable navigation planning. We characterize the long tail based on model performance and the distribution of perception-motion patterns, and employ reflection-based analysis to diagnose recurring failure modes. Experiments on web-video data and real-world urban navigation tasks demonstrate effective knowledge transfer from unconstrained videos and reveal coherent long-tail structures beyond aggregate navigation performance.

Urban NavigationVLA

Bingyi Xia, Han Bao, Zhewei Chen · 323 views · Aug 17, 2026

VioLA: Learning Generalist Humanoid Control Policies from Human Data

VioLA is a generalist humanoid policy that predicts body and hand motion latents instead of joint commands. Frozen SONIC-based encoders map human and robot demonstrations into the same action space, so 93.2% of its 140.6M training frames come from human recordings. On a real Unitree G1 it reaches 100% locomotion and 88.6% manipulation success with no task-specific fine-tuning.

HumanoidVLA

Mert Albaba, Jens Beißwenger, Anna Manasyan · 449 views · Oct 8, 2026

PhoneBot: A Low-Cost Open Humanoid Robot Platform Reusing Smartphones

PhoneBot is a roughly 400-dollar open-source 13-DoF bipedal platform that reuses a commodity smartphone as its IMU, camera and onboard computer. A torque-aware DYNAMIXEL XL430 actuator model in MJX, combined with sagittal-mirrored PPO, yields balanced walking, standup and camera-heading policies that run entirely on old or prepaid Android phones.

HumanoidOpen-source hardware

Ruochen Hou, Quanyou Wang, Daniel Koh · 628 views · Oct 6, 2026

MobileVISTA: Generative Data Augmentation for Pose Generalization in Mobile Manipulation

Mobile manipulators such as humanoid robots are increasingly deployed in dynamic, unstructured environments to perform dexterous manipulation tasks. However, end-to-end manipulation policies trained to imitate demonstration data collected from a single robot pose are brittle: even centimeter-scale deviations in robot pose at deployment can drive ego-centric observations and end-effector trajectories out of the training distribution, leading to sharp drops in performance. We introduce MobileVISTA, a data generation framework that transforms demonstrations captured at canonical poses into diverse, pose-perturbed training data by jointly (1) augmenting egocentric visual observations and (2) retargeting actions to compensate for base pose changes. Unlike prior methods, which assume a camera rigidly mounted off the actuated chain or non-trivial articulated robot geometry largely out of frame, MobileVISTA targets compatibility with egocentric platforms (e.g., humanoids) where the camera is both influenced by and must observe the robot's kinematic chain as it moves. We study MobileVISTA in simulated tasks spanning humanoid and bimanual embodiments, and on a real Galaxea R1 Pro. We find policies trained on MobileVISTA-augmented data demonstrate improved robustness to previously out-of-distribution poses encountered at test time, without additional demonstration collection or a trained generative model. Additionally, we find MobileVISTA's benefit is largest on tested humanoids, where the camera rides the actuated chain and the robot fills much of the frame. Additional videos and appendix can be found on our website: https://mobilevista.github.io

Mobile ManipulationData Augmentation

Wistreich, Suzannah, Tian, Stephen, Huang, Isabella · 332 views · Oct 5, 2026

PredActor: Predictive Action Diffusion for Steerable Onboard Humanoid Control

Diffusion models provide a flexible framework for motion generation, but turning this flexibility into closed-loop humanoid control remains challenging. Hierarchical generator-tracker systems steer motion through reference trajectories, yet these references may exceed the capabilities of the downstream tracker, leaving physical feasibility and disturbance recovery largely to a separate control module. Action-only diffusion avoids this separation by directly generating executable actions, but provides no explicit future-state trajectory that can be steered toward test-time motion objectives. Joint state-action diffusion offers a natural alternative, but existing controllers often rely on privileged full-body states, while learned behavior selection and test-time motion steering remain only partially integrated. We present PredActor, a predictive action diffusion policy that unifies both steering modes in one directly executed policy using proprioception alone. Given proprioceptive history and optional task context, PredActor jointly predicts actions and an internal future-state trajectory that enables guidance: classifier-free guidance strengthens text-conditioned motion, while classifier guidance steers future states toward test-time objectives. Only actions are executed, requiring neither a motion-reference tracker nor privileged full-body states. In simulation, PredActor reaches 44 of 45 destination targets and achieves a text retrieval score of 0.539 versus 0.424 for conditional action diffusion, with similar disturbance survival. Rolling denoising and computation-preserving runtime optimizations reduce the complete callback to 16.790 ms median and 19.383 ms p95 on a Jetson Orin NX, within the 20 ms control period. Deployed on a Unitree G1, PredActor demonstrates text-conditioned motion, disturbance response, joystick control, and semantic interpolation in simulation and hardware.

Humanoid ControlDiffusion policies

Ye, Lei, Gao, Haibo, Li, Yitang · 390 views · Sep 21, 2026

Agile and Generalized Legged Locomotion via Attention-Based Neural Map Encoding

Achieving agile and generalized legged locomotion across terrains requires tight integration of perception and control, especially under occlusions and sparse footholds. Existing methods have demonstrated agility on parkour courses but often rely on end-to-end sensorimotor models with limited generalization and interpretability. By contrast, methods targeting generalized locomotion typically exhibit limited agility and struggle with visual occlusions. We introduce a unified reinforcement learning (RL) framework for agile and generalized locomotion that incorporates a novel attention-based map encoder in the control policy. This encoder extracts local and global mapping features and uses attention mechanisms to focus on salient regions, producing an interpretable and generalized embedding for RL-based control. We further propose a learning-based mapping pipeline that provides fast, uncertainty-aware terrain representations robust to noise and occlusions, serving as policy inputs. It uses neural networks to convert depth observations into local elevations with uncertainties, and fuses them with odometry. The pipeline also integrates with parallel simulation so that we can train controllers with online mapping, aiding sim-to-real transfer. We validate our framework with the proposed mapping pipeline on a quadruped and a biped robot, and the resulting controllers demonstrate strong agility and generalization to unseen terrains in simulation and in real-world experiments.

Legged LocomotionReinforcement Learning

Chong Zhang, Victor Klemm, Fan Yang · 517 views · Jan 13, 2026

Beyond Data Scaling: Representation-Centric Continued Pre-training for Vision-Language-Action Models

Scaling robot data is crucial for building generalist Vision-Language-Action (VLA) models, yet robot trajectories are harder to scale than web-scale image-text data because embodied collection is costly and sparsely covers the physical world. This makes representation quality a central bottleneck: under a fixed robot-data budget, continued pre-training must turn limited trajectories into transferable visual-action knowledge rather than merely fit actions. We propose VLAct, a VLA-oriented VLM backbone trained on broad, heterogeneous, multi-embodiment robot data before task-specific fine-tuning. VLAct preserves the broad VLM prior and encourages shared action semantics across embodiments through VLM-prior preservation, multi-head continuous action co-supervision, and a partially unified cross-embodiment action layout, while allowing task-specific action heads during fine-tuning. Across simulation, real-world, and unseen-embodiment transfer, VLAct consistently improves downstream performance under fixed fine-tuning protocols. On LIBERO-Plus and RoboTwin 2.0, VLAct surpasses industrial VLA systems including ABot-M0 and LingBot-VLA, achieving success rates of 82.6% and 92.5%. On RoboDojo, VLAct ranks sixth among all policies by success rate and outperforms all explicitly designated world-action model (WAM) entries on both metrics. Most notably, on RoboCasa-GR1, an unseen humanoid embodiment, VLAct using only 20% of downstream trajectories outperforms the full-data GR00T-N1.6 baseline. These results are obtained using fully open-source data and only a 16-GPU training setup, showing that representation-centric continued pre-training can deliver highly competitive performance under a modest compute budget and is an important independent axis of VLA progress beyond data scaling.

VLAContinued Pre-training

Senqiao Yang, Chengyao Wang, Yuxin Chen · 401 views · Aug 27, 2026

HuMBLE: Human Motion-Driven Behavior Learning for Embodied Locomotion

Despite recent advances in humanoid locomotion, controllers optimized for command tracking and robustness tend to produce mechanical gaits, whereas controllers tied to human motion data often fail to generalize to commands outside the data distribution. This work introduces a learning framework that balances these competing objectives to synthesize real-time steerable, robust, and biomimetic locomotion policies from human data. Using an in-house curated locomotion dataset covering diverse speeds and directions, we first learn a natural locomotion prior policy through a teacher-student distillation process. Specifically, we train a full-body reference-conditioned policy with Reinforcement Learning (RL), then distill it into a lightweight prior policy conditioned solely on proprioception and a planar torso-velocity steering command. Next, we fine-tune the prior policy with multi-task RL to expand command coverage and robustness beyond the data distribution, pairing a goal-conditioned task that tracks arbitrary commands with a reference-guided task that tracks the human data as an explicit style regularizer. We validate our framework on three humanoid robots: the Boston Dynamics Atlas R1, Atlas D1, and Unitree G1. Experimental results demonstrate robust performance across real-world scenarios, including direct user-controlled locomotion in indoor and outdoor environments, and integration as the locomotion layer within hierarchical control stacks. Benchmarks against Tabula Rasa RL policies trained without human data and ablation studies confirm that our framework yields a lightweight, deployable policy that reconstructs coordinated whole-body behavior from a steering command, retaining the human gait characteristics while remaining robust and fully steerable.

Humanoid LocomotionMotion Imitation

Mike Zhang, Dongho Kang, Kevin Bergamin · 759 views · Oct 7, 2026

EgoLAP: Learning from Egocentric Human Data through Language-Action Reasoning

Egocentric human data offer a path to scaling robot learning beyond costly robot demonstrations, yet the embodiment gap makes raw human trajectories a poor supervisory target for control. Our key insight is that, although low-level actions are embodiment-specific, their underlying motion intent can capture task-relevant structure that transfers across humans and robots. We introduce EgoLAP, a VLA pre-training framework that jointly learns from human and robot trajectories through a shared language-based action chain-of-thought. EgoLAP expresses motion intent as structured, temporally abstracted language actions and pairs them with motion-level reasoning grounded in scene geometry, physics, and object affordances. Across extensive real-world and simulated experiments, EgoLAP transfers human experience to robot control more effectively than alternative action representations and reaches 80.1% mean real-world task progress, a 2.3x performance gain over alternative action representations. Motion-level reasoning also outperforms a composite reasoning format that combines subtask, object-box, and visual-trace reasoning.

VLAManipulation

Lihan Zha, Shresth Grover, Tenny Yin · 614 views · Oct 6, 2026

As an Amazon Associate, we earn from qualifying purchases.