arXiv subjects

Andrew Choi

Publications and source records attributed to Andrew Choi.

15 recordsLinked to original sources

arXiv2026

EmbodiedGen V2: An Agentic, Simulation-Ready 3D World Engine for Embodied AI

We present EmbodiedGen V2, a generative 3D world engine for building executable policy-ready environments for embodied intelligence. Sim-ready 3D asset generation has advanced rapidly, yet assembling such assets into policy-ready task environments remains largely manual, limiting scalable closed-loop learning. EmbodiedGen V2 addresses this gap through a unified sim-ready representation that connects cross-simulator assets, interaction affordances, task-driven worlds, large-scale multi-room scenes, and stateful Vibe Coding into a generative, editable, and reusable simulation pipeline. The generated environments support manipulation, navigation, mobile manipulation, cross-simulator deployment, and embodied policy training. In evaluation, the asset pipeline achieves 96.5% human acceptance and 98.6% collision success, and 83.3% of task-driven worlds are directly usable for downstream simulation without manual modification. Online reinforcement learning with generated environments further improves simulation success from 9.7% to 79.8%, and transfers to real robots with task success increasing from 21.7% to 75.0%. These results establish EmbodiedGen V2 as scalable simulation infrastructure for training, evaluating, and deploying embodied policies.

cs.RO↗

arXiv2026

RankQ: Offline-to-Online Reinforcement Learning via Self-Supervised Action Ranking

Offline-to-online reinforcement learning (RL) improves sample efficiency by leveraging pre-collected datasets prior to online interaction. A key challenge, however, is learning an accurate critic in large state--action spaces with limited dataset coverage. To mitigate harmful updates from value overestimation, prior methods impose pessimism by down-weighting out-of-distribution (OOD) actions relative to dataset actions. While effective, this essentially acts as a behavior cloning anchor and can hinder downstream online policy improvement when dataset actions are suboptimal. We propose RankQ, an offline-to-online Q-learning objective that augments temporal-difference learning with a self-supervised multi-term ranking loss to enforce structured action ordering. By learning relative action preferences rather than uniformly penalizing unseen actions, RankQ shapes the Q-function such that action gradients are directed toward higher-quality behaviors. Across sparse reward D4RL benchmarks, RankQ achieves performance competitive with or superior to seven prior methods. In vision-based robot learning, RankQ enables effective offline-to-online fine-tuning of a pretrained vision-language-action (VLA) model in a low-data regime, achieving on average a 42.7% higher simulation success rate than the next best method. In a high-data setting, RankQ improves simulation performance by 13.7% over the next best method and achieves strong sim-to-real transfer, increasing real-world cube stacking success from 43.1% to 88.9% relative to the VLA's initial performance.

cs.AI↗

arXiv2026

Scaling Sim-to-Real Reinforcement Learning for Robot VLAs with Generative 3D Worlds

The strong performance of large vision-language models (VLMs) trained with reinforcement learning (RL) has motivated similar approaches for fine-tuning vision-language-action (VLA) models in robotics. Many recent works fine-tune VLAs directly in the real world to avoid addressing the sim-to-real gap. While real-world RL circumvents sim-to-real issues, it inherently limits the generality of the resulting VLA, as scaling scene and object diversity in the physical world is prohibitively difficult. This leads to the paradoxical outcome of transforming a broadly pretrained model into an overfitted, scene-specific policy. Training in simulation can instead provide access to diverse scenes, but designing those scenes is also costly. In this work, we show that VLAs can be RL fine-tuned without sacrificing generality and with reduced labor by leveraging 3D world generative models. Using these models together with a language-driven scene designer, we generate hundreds of diverse interactive scenes containing unique objects and backgrounds, enabling scalable and highly parallel policy learning. Starting from a pretrained imitation baseline, our approach increases simulation success from 9.7% to 79.8% while achieving a 1.25$\times$ speedup in task completion time. We further demonstrate successful sim-to-real transfer enabled by the quality of the generated digital twins together with domain randomization, improving real-world success from 21.7% to 75% and achieving a 1.13$\times$ speedup. Finally, we further highlight the benefits of leveraging the effectively unlimited data from 3D world generative models through an ablation study showing that increasing scene diversity directly improves zero-shot generalization.

cs.RO↗

arXiv2025

Rapidly Learning Soft Robot Control via Implicit Time-Stepping

With the explosive growth of rigid-body simulators, policy learning in simulation has become the de facto standard for most rigid morphologies. In contrast, soft robotic simulation frameworks remain scarce and are seldom adopted by the soft robotics community. This gap stems partly from the lack of easy-to-use, general-purpose frameworks and partly from the high computational cost of accurately simulating continuum mechanics, which often renders policy learning infeasible. In this work, we demonstrate that rapid soft robot policy learning is indeed achievable via implicit time-stepping. Our simulator of choice, DisMech, is a general-purpose, fully implicit soft-body simulator capable of handling both soft dynamics and frictional contact. We further introduce delta natural curvature control, a method analogous to delta joint position control in rigid manipulators, providing an intuitive and effective means of enacting control for soft robot learning. To highlight the benefits of implicit time-stepping and delta curvature control, we conduct extensive comparisons across four diverse soft manipulator tasks against one of the most widely used soft-body frameworks, Elastica. With implicit time-stepping, parallel stepping of 500 environments achieves up to 6x faster speeds for non-contact cases and up to 40x faster for contact-rich scenarios. Finally, a comprehensive sim-to-sim gap evaluation--training policies in one simulator and evaluating them in another--demonstrates that implicit time-stepping provides a rare free lunch: dramatic speedups achieved without sacrificing accuracy.

cs.RO↗

arXiv2025

Discrete Differential Geometry for Simulating Nonlinear Behaviors of Flexible Systems: A Survey

Flexible slender structures such as rods, ribbons, plates, and shells exhibit extreme nonlinear responses bending, twisting, buckling, wrinkling, and self contact, that defy conventional simulation frameworks. Discrete Differential Geometry (DDG) has emerged as a geometry first, structure preserving paradigm for modeling such behaviors. Unlike finite element or mass spring methods, DDG discretizes geometry rather than governing equations, allowing curvature, twist, and strain to be defined directly on meshes. This approach yields robust large deformation dynamics, accurate handling of contact, and differentiability essential for inverse design and learning based control. This review consolidates the rapidly expanding landscape of DDG models across 1D and 2D systems, including discrete elastic rods, ribbons, plates, and shells, as well as multiphysics extensions to contact, magnetic actuation, and fluid structure interaction. We synthesize applications spanning mechanics of nonlinear instabilities, biological morphogenesis, functional structures and devices, and robotics from manipulation to soft machines. Compared with established approaches, DDG offers a unique balance of geometric fidelity, computational efficiency, and algorithmic differentiability, bridging continuum rigor with real time, contact rich performance. We conclude by outlining opportunities for multiphysics coupling, hybrid physics data pipelines, and scalable GPU accelerated solvers, and by emphasizing DDG role in enabling digital twins, sim to real transfer, and intelligent design of next generation flexible systems.

cond-mat.soft↗

arXiv2025

Learning Multi-Stage Pick-and-Place with a Legged Mobile Manipulator

Quadruped-based mobile manipulation presents significant challenges in robotics due to the diversity of required skills, the extended task horizon, and partial observability. After presenting a multi-stage pick-and-place task as a succinct yet sufficiently rich setup that captures key desiderata for quadruped-based mobile manipulation, we propose an approach that can train a visuo-motor policy entirely in simulation, and achieve nearly 80\% success in the real world. The policy efficiently performs search, approach, grasp, transport, and drop into actions, with emerged behaviors such as re-grasping and task chaining. We conduct an extensive set of real-world experiments with ablation studies highlighting key techniques for efficient training and effective sim-to-real transfer. Additional experiments demonstrate deployment across a variety of indoor and outdoor environments. Demo videos and additional resources are available on the project page: https://horizonrobotics.github.io/gail/SLIM.

cs.RO↗

arXiv2025

SLIM: Sim-to-Real Legged Instructive Manipulation via Long-Horizon Visuomotor Learning

We present a low-cost legged mobile manipulation system that solves long-horizon real-world tasks, trained by reinforcement learning purely in simulation. This system is made possible by 1) a hierarchical design of a high-level policy for visual-mobile manipulation following task instructions, and a low-level quadruped locomotion policy, 2) a teacher and student training pipeline for the high level, which trains a teacher to tackle long-horizon tasks using privileged task decomposition and target object information, and further trains a student for visual-mobile manipulation via RL guided by the teacher's behavior, and 3) a suite of techniques for minimizing the sim-to-real gap. In contrast to many previous works that use high-end equipments, our system demonstrates effective performance with more accessible hardware -- specifically, a Unitree Go1 quadruped, a WidowX-250S arm, and a single wrist-mounted RGB camera -- despite the increased challenges of sim-to-real transfer. Trained fully in simulation, a single policy autonomously solves long-horizon tasks involving search, move to, grasp, transport, and drop into, achieving nearly 80% real-world success. This performance is comparable to that of expert human teleoperation on the same tasks while the robot is more efficient, operating at about 1.5x the speed of the teleoperation. Finally, we perform extensive ablations on key techniques for efficient RL training and effective sim-to-real transfer, and demonstrate effective deployment across diverse indoor and outdoor scenes under various lighting conditions.

cs.RO↗

arXiv2024

Integrating Deep Metric Learning with Coreset for Active Learning in 3D Segmentation

Deep learning has seen remarkable advancements in machine learning, yet it often demands extensive annotated data. Tasks like 3D semantic segmentation impose a substantial annotation burden, especially in domains like medicine, where expert annotations drive up the cost. Active learning (AL) holds great potential to alleviate this annotation burden in 3D medical segmentation. The majority of existing AL methods, however, are not tailored to the medical domain. While weakly-supervised methods have been explored to reduce annotation burden, the fusion of AL with weak supervision remains unexplored, despite its potential to significantly reduce annotation costs. Additionally, there is little focus on slice-based AL for 3D segmentation, which can also significantly reduce costs in comparison to conventional volume-based AL. This paper introduces a novel metric learning method for Coreset to perform slice-based active learning in 3D medical segmentation. By merging contrastive learning with inherent data groupings in medical imaging, we learn a metric that emphasizes the relevant differences in samples for training 3D medical segmentation models. We perform comprehensive evaluations using both weak and full annotations across four datasets (medical and non-medical). Our findings demonstrate that our approach surpasses existing active learning techniques on both weak and full annotations and obtains superior performance with low-annotation budgets which is crucial in medical imaging. Source code for this project is available in the supplementary materials and on GitHub: https://github.com/arvindmvepa/al-seg.

cs.CV↗

arXiv2024

LatentExplainer: Explaining Latent Representations in Deep Generative Models with Multimodal Large Language Models

Deep generative models like VAEs and diffusion models have advanced various generation tasks by leveraging latent variables to learn data distributions and generate high-quality samples. Despite the field of explainable AI making strides in interpreting machine learning models, understanding latent variables in generative models remains challenging. This paper introduces LatentExplainer, a framework for automatically generating semantically meaningful explanations of latent variables in deep generative models. LatentExplainer tackles three main challenges: inferring the meaning of latent variables, aligning explanations with inductive biases, and handling varying degrees of explainability. Our approach perturbs latent variables, interprets changes in generated data, and uses multimodal large language models (MLLMs) to produce human-understandable explanations. We evaluate our proposed method on several real-world and synthetic datasets, and the results demonstrate superior performance in generating high-quality explanations for latent variables. The results highlight the effectiveness of incorporating inductive biases and uncertainty quantification, significantly enhancing model interpretability.

cs.LG↗

arXiv2023

DisMech: A Discrete Differential Geometry-based Physical Simulator for Soft Robots and Structures

Fast, accurate, and generalizable simulations are a key enabler of modern advances in robot design and control. However, existing simulation frameworks in robotics either model rigid environments and mechanisms only, or if they include flexible or soft structures, suffer significantly in one or more of these performance areas. To close this "sim2real" gap, we introduce DisMech, a simulation environment that models highly dynamic motions of rod-like soft continuum robots and structures, quickly and accurately, with arbitrary connections between them. Our methodology combines a fully implicit discrete differential geometry-based physics solver with fast and accurate contact handling, all in an intuitive software interface. Crucially, we propose a gradient descent approach to easily map the motions of hardware robot prototypes to control inputs in DisMech. We validate DisMech through several highly-nuanced soft robot simulations while demonstrating an order of magnitude speed increase over previous state of the art. Our real2sim validation shows high physical accuracy versus hardware, even with complicated soft actuation mechanisms such as shape memory alloy wires. With its low computational cost, physical accuracy, and ease of use, DisMech can accelerate translation of sim-based control for both soft robotics and deformable object manipulation.

cs.RO↗

arXiv2023

Sim2Real Neural Controllers for Physics-based Robotic Deployment of Deformable Linear Objects

Deformable linear objects (DLOs), such as rods, cables, and ropes, play important roles in daily life. However, manipulation of DLOs is challenging as large geometrically nonlinear deformations may occur during the manipulation process. This problem is made even more difficult as the different deformation modes (e.g., stretching, bending, and twisting) may result in elastic instabilities during manipulation. In this paper, we formulate a physics-guided data-driven method to solve a challenging manipulation task -- accurately deploying a DLO (an elastic rod) onto a rigid substrate along various prescribed patterns. Our framework combines machine learning, scaling analysis, and physical simulations to develop a physics-based neural controller for deployment. We explore the complex interplay between the gravitational and elastic energies of the manipulated DLO and obtain a control method for DLO deployment that is robust against friction and material properties. Out of the numerous geometrical and material properties of the rod and substrate, we show that only three non-dimensional parameters are needed to describe the deployment process with physical analysis. Therefore, the essence of the controlling law for the manipulation task can be constructed with a low-dimensional model, drastically increasing the computation speed. The effectiveness of our optimal control scheme is shown through a comprehensive robotic case study comparing against a heuristic control method for deploying rods for a wide variety of patterns. In addition to this, we also showcase the practicality of our control scheme by having a robot accomplish challenging high-level tasks such as mimicking human handwriting, cable placement, and tying knots.

cs.RO↗

arXiv2023

mBEST: Realtime Deformable Linear Object Detection Through Minimal Bending Energy Skeleton Pixel Traversals

Robotic manipulation of deformable materials is a challenging task that often requires realtime visual feedback. This is especially true for deformable linear objects (DLOs) or "rods", whose slender and flexible structures make proper tracking and detection nontrivial. To address this challenge, we present mBEST, a robust algorithm for the realtime detection of DLOs that is capable of producing an ordered pixel sequence of each DLO's centerline along with segmentation masks. Our algorithm obtains a binary mask of the DLOs and then thins it to produce a skeleton pixel representation. After refining the skeleton to ensure topological correctness, the pixels are traversed to generate paths along each unique DLO. At the core of our algorithm, we postulate that intersections can be robustly handled by choosing the combination of paths that minimizes the cumulative bending energy of the DLO(s). We show that this simple and intuitive formulation outperforms the state-of-the-art methods for detecting DLOs with large numbers of sporadic crossings ranging from curvatures with high variance to nearly-parallel configurations. Furthermore, our method achieves a significant performance improvement of approximately 50% faster runtime and better scaling over the state of the art.

cs.RO↗

arXiv2023

Learning Neural Force Manifolds for Sim2Real Robotic Symmetrical Paper Folding

Robotic manipulation of slender objects is challenging, especially when the induced deformations are large and nonlinear. Traditionally, learning-based control approaches, such as imitation learning, have been used to address deformable material manipulation. These approaches lack generality and often suffer critical failure from a simple switch of material, geometric, and/or environmental (e.g., friction) properties. This article tackles a fundamental but difficult deformable manipulation task: forming a predefined fold in paper with only a single manipulator. A sim2real framework combining physically-accurate simulation and machine learning is used to train a deep neural network capable of predicting the external forces induced on the manipulated paper given a grasp position. We frame the problem using scaling analysis, resulting in a control framework robust against material and geometric changes. Path planning is then carried out over the generated ``neural force manifold'' to produce robot manipulation trajectories optimized to prevent sliding, with offline trajectory generation finishing 15$\times$ faster than previous physics-based folding methods. The inference speed of the trained model enables the incorporation of real-time visual feedback to achieve closed-loop model-predictive control. Real-world experiments demonstrate that our framework can greatly improve robotic manipulation performance compared to state-of-the-art folding strategies, even when manipulating paper objects of various materials and shapes.

cs.RO↗

arXiv2022

A Fully Implicit Method for Robust Frictional Contact Handling in Elastic Rods

Accurate frictional contact is critical in simulating the assembly of rod-like structures in the practical world, such as knots, hairs, flagella, and more. Due to their high geometric nonlinearity and elasticity, rod-on-rod contact remains a challenging problem tackled by researchers in both computational mechanics and computer graphics. Typically, frictional contact is regarded as constraints for the equations of motions of a system. Such constraints are often computed independently at every time step in a dynamic simulation, thus slowing down the simulation and possibly introducing numerical convergence issues. This paper proposes a fully implicit penalty-based frictional contact method, Implicit Contact Model (IMC), that efficiently and robustly captures accurate frictional contact responses. We showcase our algorithm's performance in achieving visually realistic results for the challenging and novel contact scenario of flagella bundling in fluid medium, a significant phenomenon in biology that motivates novel engineering applications in soft robotics. In addition to this, we offer a side-by-side comparison with Incremental Potential Contact (IPC), a state-of-the-art contact handling algorithm. We show that IMC possesses comparable performance to IPC while converging at a faster rate.

cs.GR↗

arXiv2022

Preemptive Motion Planning for Human-to-Robot Indirect Placement Handovers

As technology advances, the need for safe, efficient, and collaborative human-robot-teams has become increasingly important. One of the most fundamental collaborative tasks in any setting is the object handover. Human-to-robot handovers can take either of two approaches: (1) direct hand-to-hand or (2) indirect hand-to-placement-to-pick-up. The latter approach ensures minimal contact between the human and robot but can also result in increased idle time due to having to wait for the object to first be placed down on a surface. To minimize such idle time, the robot must preemptively predict the human intent of where the object will be placed. Furthermore, for the robot to preemptively act in any sort of productive manner, predictions and motion planning must occur in real-time. We introduce a novel prediction-planning pipeline that allows the robot to preemptively move towards the human agent's intended placement location using gaze and gestures as model inputs. In this paper, we investigate the performance and drawbacks of our early intent predictor-planner as well as the practical benefits of using such a pipeline through a human-robot case study.

cs.RO↗