arXiv subjects

Gaoyuan Wu

Publications and source records attributed to Gaoyuan Wu.

4 recordsLinked to original sources

arXiv2026

AffordTrajDP: Dynamic Affordance-Guided Visuomotor Policy Learning for Robotic Manipulation

Affordance-guided imitation learning has shown impressive performance in robotic manipulation tasks by compressing visual perception into task-specific geometric constraints (e.g., fixed contact points). However, the commonly used static affordances can become inconsistent in precision-critical tasks or under object location perturbations, leading to post-contact trajectory drift. To address this issue, we propose AffordTrajDP, a dynamic framework that constructs affordance trajectories via object-centric temporal propagation to guide the progressive manipulation process. Specifically, given an RGB-D observation, our core insight is that a retrieved anchor affordance, which captures the desired contact point between the end-effector and the target object, can be propagated forward via affordance propagation, using the object's SE(3) pose as a natural propagation medium, to yield an affordance trajectory that provides temporally consistent, state-aware guidance throughout execution. AffordTrajDP achieves 70.0% average success rate on ManiSkill3, outperforming strong baselines by up to 17.8%. Real-world experiments on Galaxea A1 and UR7e robotic arms, covering StackCube, PickCup, AdapterInsertion, Ring-on-Peg, Put-in-Bowl, and USB Insertion, further validate robustness under object placement variations and appearance changes, with seen and unseen object instances evaluated on Galaxea A1, and ablations confirm the contribution of each proposed component.

cs.RO↗

arXiv2026

DockAnywhere: Data-Efficient Visuomotor Policy Learning for Mobile Manipulation via Novel Demonstration Generation

Mobile manipulation is a fundamental capability that enables robots to interact in expansive environments such as homes and factories. Most existing approaches follow a two-stage paradigm, where the robot first navigates to a docking point and then performs fixed-base manipulation using powerful visuomotor policies. However, real-world mobile manipulation often suffers from the view generalization problem due to shifts of docking points. To address this issue, we propose a novel low-cost demonstration generation framework named DockAnywhere, which improves viewpoint generalization under docking variability by lifting a single demonstration to diverse feasible docking configurations. Specifically, DockAnywhere lifts a trajectory to any feasible docking points by decoupling docking-dependent base motions from contact-rich manipulation skills that remain invariant across viewpoints. Feasible docking proposals are sampled under feasibility constraints, and corresponding trajectories are generated via structure-preserving augmentation. Visual observations are synthesized in 3D space by representing the robot and objects as point clouds and applying point-level spatial editing to ensure the consistency of observation and action across viewpoints. Extensive experiments on ManiSkill and real-world platforms demonstrate that DockAnywhere substantially improves policy success rates and easily generalizes to novel viewpoints from unseen docking points during training, significantly enhancing the generalization capability of mobile manipulation policy in real-world deployment.

cs.RO↗

arXiv2024

JAX-SSO: Differentiable Finite Element Analysis Solver for Structural Optimization and Seamless Integration with Neural Networks

Differentiable numerical simulations of physical systems have gained rising attention in the past few years with the development of automatic differentiation tools. This paper presents JAX-SSO, a differentiable finite element analysis solver built with JAX, Google's high-performance computing library, to assist efficient structural design in the built environment. With the adjoint method and automatic differentiation feature, JAX-SSO can efficiently evaluate gradients of physical quantities in an automatic way, enabling accurate sensitivity calculation in structural optimization problems. Written in Python and JAX, JAX-SSO is naturally within the machine learning ecosystem so it can be seamlessly integrated with neural networks to train machine learning models with inclusion of physics. Moreover, JAX-SSO supports GPU acceleration to further boost finite element analysis. Several examples are presented to showcase the capabilities and efficiency of JAX-SSO: i) shape optimization of grid-shells and continuous shells; ii) size (thickness) optimization of continuous shells; iii) simultaneous shape and topology optimization of continuous shells; and iv) training of physics-informed neural networks for structural optimization. We believe that JAX-SSO can facilitate research related to differentiable physics and machine learning to further address problems in structural and architectural design.

cs.MS↗

arXiv2022

A framework for structural shape optimization based on automatic differentiation, the adjoint method and accelerated linear algebra

Shape optimization is of great significance in structural engineering, as an efficient geometry leads to better performance of structures. However, the application of gradient-based shape optimization for structural and architectural design is limited, which is partly due to the difficulty and the complexity in gradient evaluation. In this work, an efficient framework based on automatic differentiation (AD), the adjoint method and accelerated linear algebra (XLA) is proposed to promote the implementation of gradient-based shape optimization. The framework is realized by the implementation of the high-performance computing (HPC) library JAX. We leverage AD for gradient evaluation in the sensitivity analysis stage. Compared to numerical differentiation, AD is more accurate; compared to analytical and symbolic differentiation, AD is more efficient and easier to apply. In addition, the adjoint method is used to reduce the complexity of computation of the sensitivity. The XLA feature is exploited by an efficient programming architecture that we proposed, which can boost gradient evaluation. The proposed framework also supports hardware acceleration such as GPUs. The framework is applied to the form finding of arches and different free-form gridshells: gridshell inspired by Mannheim Multihalle, four-point supported gridshell, and canopy-like structures. Two geometric descriptive methods are used: non-parametric and parametric description via Bézier surface. Non-constrained and constrained shape optimization problems are considered, where the former is solved by gradient descent and the latter is solved by sequential quadratic programming (SQP). Through these examples, the proposed framework is shown to be able to provide structural engineers with a more efficient tool for shape optimization, enabling better design for the built environment.

cs.MS↗