arXiv subjects

Guanxiong Chen

Publications and source records attributed to Guanxiong Chen.

10 recordsLinked to original sources

arXiv2026

WorldAgent: Verification-Guided Agentic Physical World Construction

Constructing complex physical worlds from language requires coordinating extensive 3D environments, detailed structures and objects at different spatial scales, and interacting physical processes under both stated goals and implicit physical constraints. We present WorldAgent, an agentic framework for verification-guided physical world construction from a single natural-language prompt, without iterative user debugging. A world construction layer expands the prompt into a structured world specification and uses physical knowledge to build scenes and run numerical simulations. After every step, a verification layer inspects scene geometry and simulation states alongside rendered views. Failed checks guide automatic revisions to the specification and re-execution of the affected steps. Accepted worlds pass the required checks and remain editable for further inspection and resimulation. We introduce AgenticSimBench, on which WorldAgent achieves the best scores among the evaluated agent-based methods on five of seven metrics. In a 26-participant user study, it receives the highest mean ratings across all four criteria.

cs.AI↗

arXiv2026

DiagGen: Agentic Generation of Deformable Assets with Sim-based Diagnostics for Robotic Simulation

While simulation-ready deformable assets are essential for in-silico robotic manipulation tasks, existing generation frameworks typically assess physical plausibility after generation, leaving an object's simulated response unused as feedback for repairing upstream errors. We present DiagGen, an agentic framework that turns a single in-the-wild image into a simulation-ready deformable asset through a generate--simulate--diagnose--refine loop. DiagGen constructs part-aware geometry and material parameters, then uses a VLM (vision-language model)-based agent to select semantically informative regions, probe them in a physics simulator, observe material responses, and route evidence-backed repair cues to the responsible generation stage. Experiments on 40 assets show that diagnostics provides useful repair cues and can moderately improve the quality of generated deformable assets. Finally, we show that unlike assets generated from visual foundation models which may not be simulatable, DiagGen-generated deformables can be directly dropped into a high-fidelity physical simulator for the planning and simulation of contact-rich pick-and-place tasks. The project's website is https://diaggen.github.io/.

cs.RO↗

arXiv2026

Agentic Real2Sim: Physics-based World Modeling with Vision-Language Agents

Real-to-sim conversion for robotic interaction with objects remains labor-intensive because it requires more than visual reconstruction: a streamlined real2sim process must recover scene geometries and object states, infer physical parameters, and assemble actors, objects, cameras, poses, and trajectories into a runnable physical simulation. Today this process still depends on brittle workflow glue across visual perception tools and simulators: manual tuning of visual foundation models, mesh cleanup, coordinate frame alignments, etc. We introduce \textit{Agentic Real2Sim}, a framework for generalized physical world modeling with vision-language agents that converts a real-world recording of object-robot interaction into a simulatable episodic twin, and connects the resulting twin to downstream policy fine-tuning and evaluation. We evaluate Agentic Real2Sim on rigid-object manipulation, deformable-object interaction, and humanoid motion scenes, spanning domains that are usually handled by separate Real2Sim pipelines. The framework's agentic decisions can be driven by an open-weight VLM backend at a small fraction of the cost of frontier models, while attaining a comparable conversion success rate. The framework further supports custom scene conversion, fine-tuning of a pretrained policy with data generated from converted episodes, and works effectively as a surrogate for real-world policy evaluation. The project site, including code is available at https://agentic-real2sim.github.io.

cs.RO↗

arXiv2026

SpringTime: Learning Simulatable Models of Cloth with Spatially-varying Constitutive Properties

Materials used in real clothing exhibit remarkable complexity and spatial variation due to common processes such as stitching, hemming, dyeing, printing, padding, and bonding. Simulating these materials, for instance using finite element methods, is often computationally demanding and slow. Worse, such methods can suffer from numerical artifacts called ``membrane locking'' that makes cloth appear artificially stiff. Here we propose a general framework, called SpringTime, for learning a simple yet efficient surrogate model that captures the effects of these complex materials using only motion observations. The cloth is discretized into a mass-spring network with unknown material parameters that are learned directly from the motion data, using a novel force-and-impulse loss function. Our approach demonstrates the ability to accurately model spatially varying material properties from a variety of data sources, and immunity to membrane locking which plagues FEM-based simulations. Compared to graph-based networks and neural ODE-based architectures, our method achieves significantly faster training times, higher reconstruction accuracy, and improved generalization to novel dynamic scenarios. Codebase for the paper can be found at https://github.com/ericchen321/springtime.

cs.GR↗

arXiv2022

ROS-X-Habitat: Bridging the ROS Ecosystem with Embodied AI

We introduce ROS-X-Habitat, a software interface that bridges the AI Habitat platform for embodied learning-based agents with other robotics resources via ROS. This interface not only offers standardized communication protocols between embodied agents and simulators, but also enables physically and photorealistic simulation that benefits the training and/or testing of vision-based embodied agents. With this interface, roboticists can evaluate their own Habitat RL agents in another ROS-based simulator or use Habitat Sim v2 as the test bed for their own robotic algorithms. Through in silico experiments, we demonstrate that ROS-X-Habitat has minimal impact on the navigation performance and simulation speed of a Habitat RGBD agent; that a standard set of ROS mapping, planning and navigation tools can run in Habitat Sim v2; and that a Habitat agent can run in the standard ROS simulator Gazebo.

cs.RO↗

arXiv2021

Transport and relaxation of current-generated nonequilibrium phonons from nonlocal electronic measurements

We study phonons generated by current in a Pt nanowire, by measuring resistance of another nanowire separated from the first one by an insulating spacer. For thin spacers, the resistance varies almost linearly with current at cryogenic temperatures, while an additional quadratic contribution emerges for thicker spacers. These observations suggest a non-thermal distribution of current-generated phonons that relax via strongly nonlinear dynamical processes rather than few-phonon scattering. Our results provide insight into the nonequilibrium phonon dynamics at nanoscale, which may facilitate efficient heat management in electronic nanodevices.

cond-mat.mes-hall↗

arXiv2021

Exchange bias without directional anisotropy in Permalloy/CoO bilayers

We utilize transverse ac susceptibility measurements to characterize magnetic anisotropy in archetypal exchange-bias bilayers of ferromagnet Permalloy (Py) and antiferromagnet CoO. Unidirectional anisotropy is observed for thin Py, but becomes negligible at larger Py thicknesses, even though the directional asymmetry of the magnetic hysteresis loop remains significant. Additional magnetoelectronic measurements, magneto-optical imaging, as well as micromagnetic simulations show that these surprising behaviors are likely associated with asymmetry of spin flop distribution created in CoO during Py magnetization reversal, which facilitates the rotation of the latter back into its field-cooled direction. Our findings suggest new possibilities for efficient realization of multistable nanomagnetic systems for neuromorphic applications.

cond-mat.mtrl-sci↗

arXiv2020

Ideal memristor based on viscous magnetization dynamics driven by spin torque

We show that ideal memristors - devices whose resistance is proportional to the charge that flows through them - can be realized using spin torque-driven viscous magnetization dynamics. The latter can be accomplished in the spin liquid state of thin-film heterostructures with frustrated exchange, where memristive response is tunable by proximity to the glass transition, while current-induced Joule heating facilitates nonvolatile operation and second-order memristive functionality beneficial for neuromorphic applications. Ideal memristive behaviors can be achieved in other systems characterized by viscous dynamics of physical, electronic, or magnetic degrees of freedom.

physics.app-ph↗

arXiv2020

Experimental demonstration and analysis of random field effects in ferromagnet/antiferromagnet bilayers

More than 30 years ago, Malozemoff (Phys. Rev. B 35, 3679 (1987)) hypothesized that exchange interaction at the interface between a ferromagnet (F) and an antiferromagnet (AF) can act as an effective random field, which can profoundly affect the magnetic properties of the system. However, until now this hypothesis has not been directly experimentally tested. We utilize magnetoelectronic measurements to analyze the effective exchange fields at Permalloy/CoO interface. Our results cannot be explained in terms of quasi-uniform effective exchange fields, but are in agreement with the random-field hypothesis of Malozemoff. The presented approach opens a new route for the quantitative analysis of effective exchange fields and anisotropies in magnetic heterostructures for memory, sensing and computing applications.

cond-mat.mtrl-sci↗

arXiv2020

Observation of anomalous non-Ohmic transport in current-driven nanostructures

Sufficiently large electric current applied to metallic nanostructures can bring them far out-of-equilibrium, resulting in non-Ohmic behaviors characterized by current-dependent resistance. We experimentally demonstrate a linear dependence of resistance on current in microscopic thin-film metallic wires at cryogenic temperatures, and show that our results are inconsistent with common non-Ohmic mechanisms such as Joule heating. As the temperature is increased, the linear dependence becomes smoothed out, resulting in the crossover to behaviors consistent with Joule heating. A plausible explanation for the observed behaviors is the strongly non-equilibrium distribution of phonons generated by the current. Analysis based on this interpretation suggests that the observed anomalous current-dependent resistance can provide information about phonon transport and electron-phonon interaction at nanoscale. The ability to control the properties of phonons generated by current can lead to new routes for the optimization of thermal properties of electronic nanodevices.

cond-mat.mes-hall↗