Skip to content

Technical Directions

VLA Latest coverage

Robotics coverage relating to vla.

414 selected reportsLast 30 days 385 recordsAll coverage 832 records

updated

Signal archive — Page 20

2026-10-16Fri381–400 records
  1. Chinese robotics — Robot learning research82

    GPT-6-Astra's Embodied Intelligence Put to the Test, Traditional Benchmarks Score Perfectly, Real Robots Hit Snags

    This research paper evaluates GPT-6-Astra's embodied intelligence capabilities across 12 simulated benchmarks and 8 real-world robot platforms. While it achieves high scores in structured tasks like LIBERO-Long++ and SIMPLE, real-world execution reveals significant limitations, including safety issues, mechanical load constraints, and failure to handle complex, unstructured tasks. The study highlights the gap between simulation and physical deployment, and proposes improvements through better execution systems and task-specific adaptations.

  2. arXiv Robotics — research abstracts75

    X-Fly: Learning to Reach and Stop in Full-Chunk Aerial Vision-Language Navigation

    This research presents X-Fly, a staged training framework for full-chunk aerial vision-language navigation. By executing complete predicted action chunks before replanning, the method reduces model inference time from 195 ms to 42 ms per waypoint and improves recovery from non-expert states. The approach combines Recovery DAgger labeling with Flow-GRPO refinement to jointly optimize a chunked flow controller and a Bernoulli STOP policy. On the OpenFly test-seen dataset, X-Fly achieves a 74.18% success rate compared to 16.45% for full-chunk behavior cloning and 32.04% for DreamFly.

  3. arXiv Robotics — research abstracts75

    A Phenomenology of Neural Geometry in Vision-Language-Action Models

    This research paper investigates the internal geometric structure of vision-language-action (VLA) policies used for robotic manipulation. By analyzing activation patterns through controlled sweeps and linear probes, the authors establish a taxonomy of seven task-relevant concepts. The findings reveal that these concepts are predominantly realized as multi-dimensional shapes influenced by physical constraints, such as wrist orientation forming an open arc due to joint limits rather than a full circle. The study confirms these geometric realizations across different VLA models on real-world robot data and demonstrates actionable methods for steering along these specific geometries to influence model behavior.

2026-10-15Thu
  1. arXiv Robotics — research abstracts78

    LEGS: Fine-Tuning Teleop-Free VLAs for Humanoid Loco-manipulation in an Embodied Gaussian Splatting World

    This research presents LEGS, a hybrid simulator that combines a mesh foreground with a photorealistic 3D Gaussian Splatting background to train vision-language-action (VLA) policies for humanoid loco-manipulation without human teleoperation. The method uses a procedural motion-primitive generator to synthesize labeled demonstrations and a two-stage color calibration to align renders with the deployment camera. Policies trained on LEGS data match or exceed those trained on human teleoperation demos and outperform mesh-only baselines. The approach enables cost-effective re-rendering of demonstrations for new scenes and objects.

  2. arXiv Robotics — research abstracts82

    Robota-Embed: Unified VLA Embeddings for Robot In-Context Learning and Evolution

    Robota-Embed is the first vision-language-action (VLA) embedding model that jointly encodes task, execution stage, and future continuation for robots. It achieves 47.33% clip-level and 81.77% episode-level R@1, outperforming action-free Gemini Embedding 2. The model supports training-free and training-based policy conditioning, improving success rates in simulated and real-world tasks, and enables task-local context evolution with reduced policy token usage.

2026-10-14Wed
  1. arXiv Robotics — research abstracts78

    Think Like a World Model, Act Like a VLA: Distilling World-Model Representations into Compact Robot Policies

    This research proposes a method to distill world-model representations into compact robot policies by aligning student policies with a frozen world model during training. The approach achieves performance improvements on tasks like LIBERO and RoboCasa-GR1, with a 0.8B student reaching 97.9% on LIBERO and improving from 48.2% to 50.5% on RoboCasa-GR1. The method is also effective on real hardware, demonstrating broad representational prior rather than fragile network alignment.

  2. arXiv Robotics — research abstracts82

    RoboGaze: Evaluating Robot World Models via Structured Vision-Language Analysis

    RoboGaze is a training-free multi-agent VLM evaluator that provides temporally localized diagnostic reports for robot world models. It outperforms existing baselines, including an agentic detector, and improves accuracy in detecting failures in synthetic manipulation videos, achieving 85% mean clean-clip accuracy on the strongest backbone.

2026-10-13Tue
  1. Chinese robotics — Robot learning research75

    MagicAtom's EchoPolicy-0 and MagicBrain Frameworks Achieve Top Performance in MolmoSpaces Benchmark

    MagicAtom's EchoPolicy-0 framework achieved a 75.4% average success rate in MolmoSpaces Combined benchmark, ranking first globally, while MagicBrain achieved 72.4% with a 94.4% grasp success rate. Both frameworks address challenges in VLA by integrating visual language models with geometric tools and action policies, improving task execution and generalization in complex environments.

  2. Chinese robotics — Robot learning research68

    MagicAtom's EchoPolicy-0 and MagicBrain Frameworks Achieve Top Performance in MolmoSpaces Benchmark

    MagicAtom's EchoPolicy-0 framework achieved a 75.4% average success rate in MolmoSpaces Combined benchmark, ranking first globally, while MagicBrain achieved 72.4% with a 94.4% grasp success rate. Both frameworks address the challenge of aligning language instructions with visual perception in embodied AI, with EchoPolicy-0 focusing on trajectory selection and MagicBrain using visual guidance for task alignment.

2026-10-12Mon
  1. arXiv Robotics — research abstracts75

    SmolWAM: A Latent World-Action Model for Robotic Manipulation on Edge Devices

    This paper introduces SmolWAM, a compact latent world-action model derived from the 2B-parameter Cosmos Policy using single-frame encoding, cached text features with preserved attention sinks, depth pruning, and token compaction. With 12 of 28 transformer blocks and 1,008 tokens, SmolWAM predicts actions and future latents in one rectified-flow step. After 3.6 A100 GPU-hours of training, it achieves 95.5% success over 400 LIBERO episodes at 57ms, 2.3TFLOP, and 2.4GiB per 16-action chunk on an A100: 11x lower latency and 3.2x lower memory than the five-step teacher. Its world head predicts scene motion; joint training improves success by 6.2 percentage points over an action-only ablation.

2026-10-10Sat
  1. AI Era — Robotics on WeChat78

    Hangzhou's Embodied Robot Rivals Figure, Demonstrates Full-Body Autonomous Clothing Folding

    The article describes how the WR1 robot, developed by West Lake Robotics, successfully performs complex household tasks such as folding clothes and managing multi-scene operations. It highlights the use of a dual pre-training system, world models, and a general-purpose action execution model to achieve seamless coordination between perception, planning, and execution. The robot's ability to handle flexible objects and maintain task continuity across different environments is emphasized as a significant breakthrough in embodied AI.

2026-10-09Fri
  1. LeRobot — Model repository updates68

    LingBot-VLA 2.0 Base Model for LeRobot

    LingBot-VLA 2.0 (base) is a pre-trained policy for visual-language agents, adapted for LeRobot. It uses a Qwen3-VL-4B backbone with a sparse-MoE Qwen2 action expert, trained via flow matching. The model is available under Apache-2.0 license and supports fine-tuning with dual-query distillation.

  3. arXiv Robotics — research abstracts82

    LM-X: Explainable Vision--Language--Action Modeling via Progress, Event, and Uncertainty Prediction

    LM-X is a new VLA foundation model that learns three directly supervised online signals—return-to-go (RTG), event-to-go (ETG), and heteroscedastic action-flow variance—to provide explainable control. Pretrained on over 20,000 hours of real-robot trajectories, LM-X achieves 74.1% success on RoboTwin2.0 tasks and outperforms GR00T N1.7 in both simulated and real-world settings. The model's signals track progress, anticipate motion, and detect high-error actions, establishing it as a transparent and effective VLA framework.

  4. arXiv Robotics — research abstracts82

    VioLA: Learning Generalist Humanoid Control Policies from Human Data

    This paper presents VioLA, a generalist humanoid policy that predicts body and hand motion latents instead of joint commands. By mapping human and robot motion into the same latent space, VioLA enables zero-shot task execution on real robots, achieving 100% success in locomotion and 88.6% in manipulation without task-specific fine-tuning. The method is validated across multiple models and datasets.

  5. arXiv Robotics — research abstracts75

    Diffusion-Steering-Based MPC with a JEPA World Action Model

    This research presents LeWAM, a bidirectional transformer designed for forward, backward, inverse dynamics and policy prediction using a decoder-free JEPA latent trained end-to-end. The authors report benefits including better state alignment via linear probes compared to regular Le World Models, reduced sensitivity to visual distractors, closed-loop performance matching flow-matching policies of equivalent size, and planning improvements when sampling raw actions or operating in the noise space of the policy head.

  6. arXiv Robotics — research abstracts82

    SAPS: Shared Autonomy for Policy Steering by Blending Teleoperation with a Pretrained VLA

    This research proposes SAPS, a lightweight framework that merges real-time human teleoperation commands with pretrained VLA actions to enhance robot manipulation reliability. Evaluated across simulated and real-world scenarios, SAPS achieves up to 77.4% hardware success, with reduced human intervention and workload, demonstrating its effectiveness for real-world deployment and assistive teleoperation.

2026-10-07Wed
  1. Robotics — Robot learning news discovery68

    Our Journey Building Vision-Language-Action Robotics

    This tutorial details the progression from imitation learning to Vision-Language-Action (VLA) models in robotics, focusing on the shift from repeating motions to understanding and executing tasks. It outlines key experiments with models like SmolVLA and π0.5, and discusses challenges such as data quality, model capacity, and simulation-to-real transfer, ultimately aiming to build robots that can see, understand, adapt, and act.

  2. Robotics — Robot learning news discovery72

    Our Journey Building Vision-Language-Action Robotics

    This article details the progression of a robotics team's work from simple imitation learning to developing Vision-Language-Action (VLA) models. It outlines challenges in adapting robotic behavior to new environments, the development of models like SmolVLA and π0.5, and the shift toward simulation-generated data for training. The focus is on building robots that can see, understand, and act autonomously in dynamic settings.

Timezone · UTC

Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).