arXiv subjects

Jidong Zhang

Publications and source records attributed to Jidong Zhang.

7 recordsLinked to original sources

arXiv2026

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

Large-scale vision--language--action (VLA) policies have advanced generalist robot control, yet most remain stimulus-to-action black boxes: actions are exposed, but their explanatory state is not. They provide no native account of three explanatory signals: task progress, the next semantic transition, or local command reliability. Prior work shows that progress and event structure aid long-horizon control and that uncertainty supports monitoring; however, such capabilities are typically added or extracted only after action pretraining. The field therefore lacks a VLA foundation model whose explanatory state is jointly pretrained with control. Drawing on biological sensorimotor organization, in which outcome-sensitive, event-segmented, and probabilistic predictions structure behavior, we introduce LM-X. LM-X learns three directly supervised online signals: return-to-go (RTG) estimates visible progress and state quality; event-to-go (ETG) predicts the action sequence to the next semantic event; and heteroscedastic action-flow variance reports local command reliability. RTG conditions ETG and both condition action generation; uncertainty is estimated inside the action expert, making explanation part of control rather than a post-hoc description. We pretrain LM-X on more than 20,000 hours of heterogeneous real-robot trajectories, including over 1,000 hours of failed rollouts. A controlled gate favors joint over post-hoc training. LM-X achieves 74.1\% success on 50 randomized-hard RoboTwin2.0 tasks and 73.5\% on seven real-robot tasks, compared with 55.4\% and 50.7\% for GR00T N1.7. Its signals track progress and regression, anticipate event-scale motion, detect high-error actions, and provide advance failure warning. These results establish LM-X as an explainable VLA foundation model that couples transparent predictive state with stronger generalist control.

cs.RO↗

arXiv2026

RoboCousin: Build Your Own Simulation Playground for Robust Bimanual Robotic Manipulation

Bimanual manipulation policies require large and diverse training datasets, yet collecting demonstrations on physical robots is expensive and difficult to scale. Simulation can generate data efficiently, but existing pipelines typically operate within closed asset libraries and predefined scenes: adding a newly observed object or environment still requires substantial effort to reconstruct geometry, specify physical and semantic properties, annotate interactions, and integrate the result into executable tasks. We present RoboCousin, an extensible simulation-based data-generation platform that turns user-provided observations into reusable assets, scenes, and expert trajectories for bimanual manipulation. Built on RoboTwin~2.0, RoboCousin converts object images into simulation-ready assets with visual and collision geometry, semantic and physical metadata, and automatically generated grasp-contact candidates. It further constructs digital cousins that vary compatible objects, backgrounds, layouts, and language instructions while preserving task-relevant affordances and spatial relations. The same asset system supports tabletop and room-level scene construction, with collision-aware base control for interaction beyond a fixed workspace. We release RoboCousin-OBD, containing more than 3,000 annotated object instances and 50 background environments, and use RoboCousin to generate over one million expert trajectories across 50 tasks. Simulation and real-robot experiments show that the automatically generated interaction annotations are comparable to curated annotations, generated assets provide effective sim-to-real supervision, and tabletop cousins can improve transfer beyond training on a single reconstructed scene. RoboCousin therefore provides a practical path for expanding both the scale and coverage of synthetic bimanual manipulation data.

cs.RO↗

arXiv2017

First-principles study of phononic thermal transport in monolayer C3N: a comparison with graphene

Very recently, a new graphene-like crystalline, hole-free, 2D-single-layer carbon nitride C3N, has been fabricated by polymerization of 2,3-diaminophenazine and used to fabricate a field-effect transistor device with an on-off current ratio reaching (Adv. Mater. 2017, 1605625). Heat dissipation plays a vital role in its practical applications, and therefore the thermal transport properties need to be explored urgently. In this paper, we perform first-principles calculations combined with phonon Boltzmann transport equation to investigate the phononic thermal transport properties of monolayer C3N, and meanwhile, a comparison with graphene is given. Our calculated intrinsic lattice thermal conductivity of C3N is 380 W/mK at room temperature, which is one order of magnitude lower than that of graphene (3550 W/mK at 300 K), but is greatly higher than many other typical 2D materials. The underlying mechanisms governing the thermal transport were thoroughly discussed and compared to graphene, including group velocities, phonon relax time, the contribution from phonon branches, phonon anharmonicity and size effect. The fundamental physics understood from this study may shed light on further studies of the newly fabricated 2D crystalline C3N sheets.

physics.comp-ph↗

arXiv2016

Biphoton statistic of quantum light generated on a silicon chip

We demonstrate a silicon-chip biphoton source with an unprecedented quantum cross correlation up to ${\rm g_{si}^{(2)}(0) = (2.58 \pm 0.16) \times 10^4}$. The emitted biphotons are intrinsically single-mode, with self correlations of ${\rm g_{ss}^{(2)}(0) = 1.90 \pm 0.05}$ and ${\rm g_{ii}^{(2)}(0) = 1.87 \pm 0.06}$ for signal and idler photons, respectively. We observe the waveform asymmetry of cross correlation between signal and idler photons and reveal the identical and non-exponential nature of self correlations of individual signal and idler photon modes, which is a nature of cavity-enhanced nonlinear optical processes. The high efficiency and high purity of the biphoton source allow us to herald single photons with a conditional self correlation $\rm g_{c}^{(2)}(0)$ as low as $\rm 0.0059 \pm 0.0014$ at a pair flux of $\rm 1.95 \times 10^5$ pairs/s, which remains below $\rm 0.026 \pm 0.001$ for a biphoton flux up to $\rm 2.93 \times 10^6$ pairs/s, with a photon preparation efficiency in the single-mode fiber up to 51%, among the best values that have ever been reported. Our work unambiguously demonstrates that silicon photonic chips are superior material and device platforms for integrated quantum photonics.

physics.optics↗

arXiv2014

Dispersion engineering of high-Q silicon microresonators via thermal oxidation

We propose and demonstrate a convenient and sensitive technique for precise engineering of group-velocity dispersion in high-Q silicon microresonators. By accurately controlling the surface-oxidation thickness of silicon microdisk resonators, we are able to precisely manage the zero-dispersion wavelength while simultaneously further improving the high optical quality of our devices, with the optical Q close to a million. The demonstrated dispersion management allows us to achieve parametric generation with precisely engineerable emission wavelengths, which shows great potential for application in integrated silicon nonlinear and quantum photonics.

physics.optics↗

arXiv2012

A silicon-chip source of bright photon-pair comb

Integrated quantum photonics relies critically on the purity, scalability, integrability, and flexibility of a photon source to support diverse quantum functionalities on a single chip. Up to date, it remains an open challenge to realize an efficient monolithic photon-pair source for on-chip application. Here we report a device on the silicon-on-insulator platform that utilizes dramatic cavity enhanced four-wave mixing in a high-Q silicon microdisk resonator. The device is able to produce high-purity photon pairs in a comb fashion, with an unprecedented spectral brightness of $6.24 \times 10^7 {\rm pair/s/mW^2/GHz}$ and photon-pair correlation with a coincidence-to-accidental ratio of $1386 \pm 278$ while pumped with a continuous-wave laser. The superior performance, together with the structural compactness and CMOS compatibility, opens up a great avenue towards quantum silicon photonics with unprecedented capability of multi-channel parallel information processing for both integrated quantum computing and long-haul quantum communication.

physics.optics↗

arXiv2012

Status of polarization control experiment at Shanghai deep ultraviolet free electron laser

A polarization control experiment by utilizing a pair of crossed undulators has been proposed for the Shanghai deep ultraviolet free electron laser test facility. Numerical simulations indicate that, with the electromagnetic phase-shifter located between the two crossed planar undulators, fully coherent radiation with 100 nJ order pulse energy, 5 picoseconds pulse length and circular polarization degree above 90% could be generated. The physical design study and the preparation status of the experiment are presented in the paper.

physics.acc-ph↗