Neue Studien zu Robotik und Physical AI, jeweils mit ihrer Bedeutung für die Branche.
Vision-Language-Action (VLA) models commonly generate robot actions token by token through Large Language Model (LLM) decoders, causing inference latency to grow with action sequence length and limiting closed-loop control frequency. We present Goal-Conditioned Vision-Predictor-Action (VPA), which replaces autoregressive decoders with a lightweight latent control loop.
Runtime failure monitors for robot policies are typically evaluated by how accurately they distinguish failed from successful executions. Yet an accurate alarm may arrive too late for recovery, while an unnecessary intervention may disrupt an execution that would otherwise succeed.
Reinforcement Learning (RL) substantially improves planning performance in autonomous-driving Vision-Language-Action (VLA) models. However, existing RL methods rely on scalar trajectory rewards that measure overall quality but fail to identify specific planning errors or provide actionable correction signals.
Vision-language-action (VLA) models and world action models (WAMs) have advanced robotic policy learning by building on pretrained vision-language models (VLMs) and video generative models (VGMs), respectively. Yet VLAs learn without explicit supervision from future world states, while WAMs that condition actions on predicted futures incur extra inference cost and let visual prediction errors propagate into actions.
Vision-language-action (VLA) policies issue sparse targets whose arrival at the robot controller is delayed and irregular, so a reasonable predicted action can still produce abrupt motion near people. Remedies are evaluated on different streams, delays and metrics.
Memory-dependent manipulation requires a policy to track execution state and recover visual evidence that is no longer observable. We propose dual-stream memory for vision-language-action policies, assigning these requirements to complementary representations.
Pre-trained robot policies, such as Vision-Language-Action (VLA) models, provide a strong foundation for generalist robotic manipulation, but adapting them to new tasks with only a few demonstrations remains challenging. Directly fine-tuning a pre-trained policy often achieves strong in-distribution performance, yet it can distort the representations learned during pre-training and weaken the model's generalization ability under out-of-distribution shifts.
Pretrained vision-language-action (VLA) models encode broad manipulation priors, yet adapting them to specific tasks requires balancing limited robot interaction, computational resources, and safety constraints. Noise space learning (NSL) adapts flow-matching-based VLAs by learning a context-conditioned distribution over noise while keeping the pretrained VLA fixed.
Actions specify what a robot should execute, but do not describe what those commands will achieve. Vision-language-action (VLA) policies nevertheless learn primarily by predicting action chunks from images, language, and state.
A user who brings a generalist manipulation policy into their home or office wants it to pick up their objects from their shelves, counters, and tables. Yet even a strong generalist like $π_{0.5}$ achieves 0% pickup success in our captured deployment scenes without adaptation, and collecting real demonstrations in every user's environment is prohibitively expensive.
Scaling generalist robotic policies is limited by the high cost of collecting embodied interaction data. Large-scale egocentric human videos offer a rich alternative, but they lack robot action labels, and existing action annotation methods are often noisy under real-world visual complexity and diverse manipulation dynamics.
Das Preprint beschreibt FastOPD, ein Framework, das große Vision-Language-Action-Modelle per On-Policy-Destillation in kompakte Studenten mit wenigen Inferenzschritten überführt. Es kombiniert eine einzelne On-Policy-Lehrerüberwachung mit Selbstkonsistenz und erreicht auf LIBERO mit zwei Schritten 81,8 % Erfolgsrate, was 84 % der Leistung von π0.5 entspricht, bei 78,1 % geringerer Latenz. Das Verfahren senkt Trainings- und Inferenzkosten deutlich und wurde auch real auf einem Roboter demonstriert.
Das Paper stellt MixVLA vor, ein als Preprint veröffentlichtes, modellagnostisches Trainingsverfahren für Vision-Language-Action-Modelle. Es zerlegt latente Repräsentationen in invariante und nicht-invariante Anteile und regularisiert letztere durch adaptives Mischen, statt sie zu verwerfen. In Simulationen verbessert MixVLA die Zero-Shot-Erfolgsraten auf LIBERO-Plus um 6,6 Prozentpunkte gegenüber OpenVLA-OFT und auf RoboTwin 2.0 um 16,4 Prozentpunkte bei der Generalisierung auf randomisierte Umgebungen; im realen Versuch steigt die Erfolgsrate von 20 % auf 60 %.
Der Preprint identifiziert Instruction-Action-Binding als Fehlermodus, bei dem VLA-Policies trotz aktiver Sprach- und Bildverarbeitung vertraute, aber falsche Trajektorien auswählen. Equivariant Counterfactual Training (ECT) ergänzt Trainingsdaten um szenenabhängige Aktionsalternativen unter derselben Anweisung und trainiert diese Paare gemeinsam. Auf LIBERO-PRO steigt der mittlere Swap-Erfolg von π0.5 von 36 auf 59 Prozent, auf CALVIN verbessert der ECT-Loss die Fünf-Aufgaben-Abschlussrate von 58 auf 76 Prozent, und auf einem echten UR5e steigt der Erfolg bei ungesehenen Positionen von 3/40 auf 35/40.
Roboter-KI-Modelle planen immer einen kurzen Block künftiger Bewegungen, und wie viele davon der Roboter ausführt, bevor er wieder hinschaut, legen Entwickler bislang meist starr fest. Dieser Preprint der Tsinghua-Universität, der Beijing Academy of Artificial Intelligence (BAAI) und weiterer Partner ergänzt ein Zusatzmodul, das diese Zahl im laufenden Betrieb wählt und die Erfolgsquoten vorhandener Modelle von Physical Intelligence und Nvidia in der Simulation und auf einem realen Zweiarmroboter ohne Nachtraining erhöht.
Vision-language-action (VLA) models have advanced robotic manipulation, but their zero-shot generalization in new tasks and environments remains limited, and their reliance on specialized training keeps them from benefiting directly from rapidly advancing general-purpose vision-language models (VLMs). In parallel, recent agentic robotic systems leverage VLMs for high-level reasoning or coding agents for robot control, but often depend on extensive external models and tools, introducing additional complexity and cost.
Visual Language Action (VLA) models offer unprecedented generalization for autonomous robots; however, their real-world deployment is frequently bottlenecked by unreliable execution and the prohibitive computational cost of fine-tuning for specific robot embodiments and tasks. To bridge this gap, we propose WayFinder, an end-to-end, closed-loop hierarchical VLA framework that circumvents the need for fine-tuning by decoupling high-level task reasoning from low-level kinematic control.
Vision-language-action and world-action models have demonstrated impressive capabilities in robotics, yet generalization to unseen tasks remains challenging. More recently, general-purpose multimodal agents have shown great potential for zero-shot robotic task solving.
Diffusion and flow-matching Vision-Language-Action (VLA) policies generate action chunks through iterative denoising, incurring substantial inference latency that severely limits real-time robotic control. Existing acceleration methods treat an action chunk as a monolithic computational unit, ignoring a crucial physical reality of receding-horizon control: actions are generated jointly but consumed sequentially, resulting in inherently heterogeneous execution urgencies.
Simulated manipulation benchmarks are the standard tool for evaluating vision-language-action (VLA) policies and the acceleration methods that reduce their inference latency for on-robot deployment. On these benchmarks, we observe that some training-free acceleration methods, which approximate the baseline policy's computation, achieve higher measured success rates than the baseline itself.
Die Studien stammen aus den Robotik-Feeds von arXiv. Solange unsere Zusammenfassung fehlt, erscheinen die ersten Zeilen des Abstracts.