Tag
PINT 提出了一种统一模型，通过可学习的预测意图令牌（Predictive Intent Tokens）协调视觉语言推理、视频世界建模与动作生成，在无需具身预训练的情况下显著提升了机器人操作的分布外泛化能力，在 RoboTwin 2.0-Hard 和 LIBERO-Plus 等基准上大幅超越 Fast-WAM 和 StarVLA。
This paper analyzes how VLA policies should choose which backbone layers form the latent interface, finding that layer fusion often underperforms the best single layer and proposing InfoNCE as a compute-efficient proxy for layer selection that cuts GPU costs 9-33x while reducing selection regret.
This paper formalizes multimodality in generative behavioral cloning and finds, through theory and extensive experiments on LIBERO, Meta-World, and synthetic benchmarks, that common simulation benchmarks are largely unimodal and that simple deterministic action transformers remain competitive with expressive generative policies such as diffusion, flow, and CVAE-based methods.
MotorMind introduces a harness that lets a general-purpose VLM control a robot directly by reasoning over observations and issuing mid-level actions with asynchronous feedback, achieving 66.7% success on LIBERO-PRO zero-shot (vs. up to 13.3% prior methods) and 95% on a real xArm6 robot without task-specific training, coding agents, or grounding tools.
The paper proposes Predictive Action Chunk Learning (PACL) for robot manipulation, which learns from mixed-quality deployment experiences using a predictive critic and diffusion actor to improve policy performance.
The paper presents World Action Agent (WAA), a multi-agent system that uses Vision-Language Models (VLMs) for robot manipulation through a visual action workspace with contact views, action rehearsal, and in-view correction, achieving state-of-the-art results on benchmarks.
DeformSmith is a framework for generating interactive, physically credible deformable assets for robot manipulation from text or images, using physics-guided hierarchical generation to improve quality and plausibility.
This paper presents CosmoH2G, a dataset and baseline method for transferring complex human hand demonstrations to robotic grippers using a two-stage framework that predicts sparse keyframes and continuous actions to handle intricate spatial movements.
StreamPI introduces a streaming multimodal temporal modeling framework for vision-language-action models, improving robot manipulation through instruction-anchored attention and randomized interval training without additional parameters.
DyPES-VLA is a cross-embodiment VLA model that learns shared dynamics priors via future prediction and uses an embodiment-specific Mixture-of-Experts action head to control robots in their native action spaces, achieving state-of-the-art results on LIBERO, RoboCasa, and RoboTwin benchmarks.
Introduces W2-VLA, a vision-language-action model for fine-grained robot manipulation that models task-conditioned future wrist observations, achieving improved manipulation performance on benchmarks and real-world tasks.
BridgeVLA++ is a memory-augmented vision-language-action framework for 3D robot manipulation that builds on BridgeVLA to add spatio-temporal memory, achieving state-of-the-art results on memory-dependent manipulation benchmarks while preserving data efficiency and generalization.
DreamTraj predicts 6-DoF object trajectories from a single RGB image and a language instruction by decoding internal video diffusion latents, eliminating the need for video, depth, or CAD models at inference. It introduces the MOVEdataset with fine-grained language-to-motion annotations and achieves state-of-the-art performance while running 4.6x faster than generate-then-extract pipelines.
πR^2 introduces a reactive real-time flow policy for robot manipulation that splits conditioning into fast and slow channels and uses a latency-adaptive flow schedule, enabling closed-loop replanning ~4x faster than baseline policies and improving success rates by up to 30%.
Grabette is an open, low-cost system for recording robot manipulation data using a handheld gripper and camera, aiming to build a shared dataset for robot learning.
Introduces B-spline Policy (BSP), which parameterizes actions as continuous B-spline curves instead of discrete fixed-rate action chunks, enabling faster and smoother manipulation on low-cost robot arms.
CLAP proposes a method to convert pretrained vision-language models (VLMs) into vision-language-action models (VLAs) by prepending natural-language action descriptions to action token sequences, preserving semantic capabilities without architectural changes. It achieves 90.8% on LIBERO and improves robustness.
The paper introduces Prompt-Driven Exploration (PDE), a method that uses a vision-language model to iteratively refine natural language prompts for reinforcement learning policies, enabling global exploration and successful policy learning even from zero-reward starts.
InternVLA-A1.5 integrates pretrained vision-language models with future prediction in latent space to enable efficient robot manipulation with compositional generalization and long-horizon execution, achieving state-of-the-art results on simulation benchmarks.
Almond Robotics' Axol robot demonstrates folding a towel.