arXiv subjects

Ding Yi

Publications and source records attributed to Ding Yi.

3 recordsLinked to original sources

arXiv2026

Evolve Vision-Language-Action Model into an Agent with On-the-fly Tool-use

This paper integrates end-to-end Visual-Language-Action (VLA) models with agentic tool-use to propose Agentic Robot with Tool-use (ART). ART is a tool-injection framework that tunes any VLA model to leverage off-the-shelf tool modules for low-level vision, high-level affordance, and embodiment enhancement. Compared to vanilla VLA models with a whole continuous action solution space, ART reduces the complexity of the action solution space through tool-use, which not only improves generalizability across different tasks but also reduces data dependency. To demonstrate the advantages (high generalizability and low data dependency) of this framework, we first built a dataset of 30K tool-use trajectories and action demonstrations, which is much smaller than those used by baseline methods. We then designed a training regimen for long-trajectory tool-use reasoning in challenging environments. Experiments show that ART achieves a 20% higher success rate than mainstream baselines on simulation and real-world tasks, such as pick-and-place in the dark at novel viewpoints. Empirical results highlight the benefits of an agent-based approach: modular tool utilization enables more efficient training, lightweight deployment, and scalable integration of new tools. This design fosters robustness, adaptability, and extensibility, paving the way for the practical deployment of VLA systems in complex real-world scenarios.

cs.RO↗

arXiv2026

Watch, Recall, Act: Always-On Robots in Concurrent Embodied Streams

An always-on robot faces an endless stream that never resets: instructions arrive and lapse, the scene changes, and its own past actions reshape what it must reason about. Today's action models are built for the opposite: a fixed instruction, no mid-task intervention, single-step reasoning. In an open-ended world a robot must watch a live stream for far-future cues, recall its own far-past actions, and act on them under dual-arm concurrency. We present ARMS (Always-on Robot in Multi-modal Streams), a deliberately simple streaming policy: a single pretrained $π$0.5 backbone augmented by three lightweight modules that turn live perception, embodied states, and the robot's own past actions into context the backbone reads before it acts. The modules update this context asynchronously, so watching and recalling never block acting and the two arms act at once. Rather than inventing new mechanisms, ARMS integrates these learned context providers with an agent-causal self-history that logs which arm did what, and when. To supervise them without extra annotation, we build ARMS Dataset, whose staged construction script itself labels every module from real dual-arm teleoperation. Trained on it, ARMS reaches 45% on the combined task against 28% for the strongest of our four main baselines, and ablations confirm the memory module, the embodied-state head, and asynchronous concurrency are each necessary.

cs.RO↗

arXiv2017

Ultrafast Epitaxial Growth of Metre-Sized Single-Crystal Graphene on Industrial Cu Foil

A foundation of the modern technology that uses single-crystal silicon has been the growth of high-quality single-crystal Si ingots with diameters up to 12 inches or larger. For many applications of graphene, large-area high-quality (ideally of single-crystal) material will be enabling. Since the first growth on copper foil a decade ago, inch-sized single-crystal graphene has been achieved. We present here the growth, in 20 minutes, of a graphene film of 5 x 50 cm2 dimension with > 99% ultra-highly oriented grains. This growth was achieved by: (i) synthesis of sub-metre-sized single-crystal Cu(111) foil as substrate; (ii) epitaxial growth of graphene islands on the Cu(111) surface; (iii) seamless merging of such graphene islands into a graphene film with high single crystallinity and (iv) the ultrafast growth of graphene film. These achievements were realized by a temperature-driven annealing technique to produce single-crystal Cu(111) from industrial polycrystalline Cu foil and the marvellous effects of a continuous oxygen supply from an adjacent oxide. The as-synthesized graphene film, with very few misoriented grains (if any), has a mobility up to ~ 23,000 cm2V-1s-1 at 4 K and room temperature sheet resistance of ~ 230 ohm/square. It is very likely that this approach can be scaled up to achieve exceptionally large and high-quality graphene films with single crystallinity, and thus realize various industrial-level applications at a low cost.

cond-mat.mtrl-sci↗