arXiv subjects

Jiayun Li

Publications and source records attributed to Jiayun Li.

At least 19 recordsLinked to original sources

arXiv2026

Globalized Constrained Stein Variational Inference for Diverse Feasible Robot Motion Planning

Robot motion planning is inherently multimodal, yet classical planners typically return only a single solution. Probabilistic formulations address this limitation by maintaining a distribution over motions, allowing the planner to reason over multiple low-cost alternatives. In robotics, however, motion samples must also satisfy strict constraints, including collision avoidance, joint limits, contact conditions, and dynamics consistency. These hard requirements make motion sampling substantially more challenging: within a limited planning budget, the ensemble must cover diverse low-cost motions while ensuring that every sample remains feasible under the relevant constraints. We propose SteinSQP (Stein Variational Sequential Quadratic Programming), a constrained Stein variational inference method for diverse feasible robot motion sampling. SteinSQP evolves an interacting particle ensemble, as in Stein variational methods, while embedding constraints directly into a kernel-space SQP subproblem. We solve the resulting constrained Stein-Newton subproblem with a GPU-friendly matrix-free primal-dual algorithm, enabling efficient batched ensemble updates. To globalize the method, we introduce an ensemble-level merit function that jointly balances objective value, constraint violation, and particle diversity. Across five constrained motion-planning tasks, SteinSQP returns fully feasible ensembles while preserving diverse motion alternatives. Compared with first-order constrained Stein baselines and serial multistart nonlinear programming, SteinSQP shows faster and more robust ensemble convergence in terms of iterations, improves particle-wise feasibility, and achieves faster batched time-to-solution on challenging robot-scale tasks.

cs.RO↗

arXiv2026

RCTEA: Richness-guided Co-training for Temporal Entity Alignment

Temporal Entity Alignment (TEA), which aims to identify equivalent entities across Temporal Knowledge Graphs (TKGs), is crucial for integrating knowledge facts from multiple sources. However, existing TEA models often fail to capture the orthogonal yet complementary effects between structural and temporal features, and typically overlook the importance of information richness, a key factor for effective message passing in neural feature encoders. To address these limitations, we propose the RCTEA framework, which jointly models both structural and temporal aspects of TKGs for entity alignment. Specifically, we design a richness-guided attention mechanism along with an adaptive weighting strategy to facilitate effective feature fusion. To ensure robust alignment despite noisy entity contexts, we introduce a dual-view neighborhood consensus algorithm that jointly refines the feature encoders to enforce local structural consistency of the predicted alignments. Extensive experiments demonstrate the superiority of RCTEA, achieving state-of-the-art performance on public TEA benchmarks.

cs.IR↗

arXiv2026

IMPACT: An Implicit Active-Set Augmented Lagrangian for Fast Contact-Implicit Trajectory Optimization

Contact-implicit trajectory optimization (CITO) has attracted growing attention as a unified framework for planning and control in contact-rich robotic tasks. Recent approaches have demonstrated promising results in manipulation and locomotion without requiring a prescribed contact-mode schedule. It is well known that the underlying mathematical programs with complementarity constraints (MPCCs) remain numerically ill-conditioned, and systematic, scalable solution strategies for CITO remain an active area of research. More efficient and principled solvers that can handle contact constraints are therefore essential to broaden the applicability of CITO. In this work, we develop an augmented-Lagrangian approach to CITO for solving MPCC-based CITO with stationarity guarantees. The method can be interpreted as identifying the implicit contact-mode branches on the fly during the trajectory optimization (TO) iterations; we call this approach IMPACT (IMPlicit contact ACtive-set Trajectory optimization). We provide an efficient C++ implementation tailored to trajectory-optimization workloads and evaluate it on the open-source CITO and contact-implicit model predictive control (CI-MPC) benchmarks. On CITO, IMPACT achieves 2.9x-70x speedups over strong baselines (geometric mean 13.8x). On CI-MPC, we show improved control quality for contact-rich trajectories on dexterous manipulation tasks in simulation. Finally, we demonstrate the proposed method on real robotic hardware on a T-shaped object pushing task.

cs.RO↗

arXiv2026

Stein Variational Ergodic Surface Coverage with SE(3) Constraints

Surface manipulation tasks require robots to generate trajectories that comprehensively cover complex 3D surfaces while maintaining precise end-effector poses. Existing ergodic trajectory optimization (TO) methods demonstrate success in coverage tasks, while struggling with point-cloud targets due to the nonconvex optimization landscapes and the inadequate handling of SE(3) constraints in sampling-as-optimization (SAO) techniques. In this work, we introduce a preconditioned SE(3) Stein Variational Gradient Descent (SVGD) approach for SAO ergodic trajectory generation. Our proposed approach comprises multiple innovations. First, we reformulate point-cloud ergodic coverage as a manifold-aware sampling problem. Second, we derive SE(3)-specific SVGD particle updates, and, third, we develop a preconditioner to accelerate TO convergence. Our sampling-based framework consistently identifies superior local optima compared to strong optimization-based and SAO baselines while preserving the SE(3) geometric structure. Experiments on a 3D point-cloud surface coverage benchmark and robotic surface drawing tasks demonstrate that our method achieves superior coverage quality with tractable computation in our setting relative to existing TO and SAO approaches, and is validated in real-world robot experiments.

cs.RO↗

arXiv2026

S3-CoT: Self-Sampled Succinct Reasoning Enables Efficient Chain-of-Thought LLMs

Large language models (LLMs) equipped with chain-of-thought (CoT) achieve strong performance and offer a window into LLM behavior. However, recent evidence suggests that improvements in CoT capabilities often come with redundant reasoning processes, motivating a key question: Can LLMs acquire a fast-thinking mode analogous to human System 1 reasoning? To explore this, our study presents a self-sampling framework based on activation steering for efficient CoT learning. Our method can induce style-aligned and variable-length reasoning traces from target LLMs themselves without any teacher guidance, thereby alleviating a central bottleneck of SFT-based methods-the scarcity of high-quality supervision data. Using filtered data by gold answers, we perform SFT for efficient CoT learning with (i) a human-like dual-cognitive system, and (ii) a progressive compression curriculum. Furthermore, we explore a self-evolution regime in which SFT is driven solely by prediction-consistent data of variable-length variants, eliminating the need for gold answers. Extensive experiments on math benchmarks, together with cross-domain generalization tests in medicine, show that our method yields stable improvements for both general and R1-style LLMs. Our data and model checkpoints can be found at https://github.com/DYR1/S3-CoT.

cs.CL↗

arXiv2026

From Latent Signals to Reflection Behavior: Tracing Meta-Cognitive Activation Trajectory in R1-Style LLMs

R1-style LLMs have attracted growing attention for their capacity for self-reflection, yet the internal mechanisms underlying such behavior remain unclear. To bridge this gap, we anchor on the onset of reflection behavior and trace its layer-wise activation trajectory. Using the logit lens to read out token-level semantics, we uncover a structured progression: (i) Latent-control layers, where an approximate linear direction encodes the semantics of thinking budget; (ii) Semantic-pivot layers, where discourse-level cues, including turning-point and summarization cues, surface and dominate the probability mass; and (iii) Behavior-overt layers, where the likelihood of reflection-behavior tokens begins to rise until they become highly likely to be sampled. Moreover, our targeted interventions uncover a causal chain across these stages: prompt-level semantics modulate the projection of activations along latent-control directions, thereby inducing competition between turning-point and summarization cues in semantic-pivot layers, which in turn regulates the sampling likelihood of reflection-behavior tokens in behavior-overt layers. Collectively, our findings suggest a human-like meta-cognitive process-progressing from latent monitoring, to discourse-level regulation, and to finally overt self-reflection. Our analysis code can be found at https://github.com/DYR1/S3-CoT.

cs.CL↗

arXiv2025

Identification with Orthogonal Basis Functions: Convergence Speed, Asymptotic Bias, and Rate-Optimal Pole Selection

This paper is concerned with performance analysis and pole selection problem in identifying linear time-invariant (LTI) systems using orthogonal basis functions (OBFs), a system identification approach that consists of solving least-squares problems and selecting poles within the OBFs. Specifically, we analyze the convergence properties and asymptotic bias of the OBF algorithm, and propose a pole selection algorithm that robustly minimizes the worst-case identification bias, with the bias measured under the $\mathcal{H}_2$ error criterion. Our results include an analytical expression for the convergence rate and an explicit bound on the asymptotic identification bias, which depends on both the true system poles and the preselected model poles. Furthermore, we demonstrate that the pole selection algorithm is asymptotically optimal, achieving the fundamental lower bound on the identification bias. The algorithm explicitly determines the model poles as the so-called Tsuji points, and the asymptotic identification bias decreases exponentially with the number of basis functions, with the rate of decrease governed by the hyperbolic Chebyshev constant. Numerical experiments validate the derived bounds and demonstrate the effectiveness of the proposed pole selection algorithm.

math.OC↗

arXiv2025

Conditional Generative Modeling of Stochastic LTI Systems: A Behavioral Approach

This paper presents a data-driven model for Linear Time-Invariant (LTI) stochastic systems by sampling from the conditional probability distribution of future outputs given past input-outputs and future inputs. It operates in a fully behavioral manner, relying solely on the current trajectory and pre-collected input-output data, without requiring explicit identification of system parameters. We refer to this model as a behavioral Conditional Generative Model (CGM). We prove the convergence of the distribution of samples generated by the CGM as the size of the trajectory library increases, with an explicit characterization of the convergence rate. Furthermore, we demonstrate that the gap between the asymptotic distribution of the proposed CGM and the true posterior distribution obtained by Kalman filter, which leverages the knowledge of all system parameters and all historical data, decreases exponentially with respect to the length of past samples. Finally, we integrate this generative model into predictive controllers for stochastic LTI systems. Numerical results verify the derived bounds and demonstrate the effectiveness of the controller equipped with the proposed behavioral CGM.

math.OC↗

arXiv2025

Finite Sample Performance Analysis of MIMO Systems Identification

This paper is concerned with the finite sample identification performance of an n dimensional discrete-time Multiple-Input Multiple-Output (MIMO) Linear Time-Invariant system, with p inputs and m outputs. We prove that the widely-used Ho-Kalman algorithm and Multivariable Output Error State Space (MOESP) algorithm are ill-conditioned for MIMO systems when n/m or n/p is large. Moreover, by analyzing the Cramer-Rao bound, we derive a fundamental limit for identifying the real and stable (or marginally stable) poles of MIMO system and prove that the sample complexity for any unbiased pole estimation algorithm to reach a certain level of accuracy explodes superpolynomially with respect to n/(pm). Numerical results are provided to illustrate the ill-conditionedness of Ho-Kalman algorithm and MOESP algorithm as well as the fundamental limit on identification.

eess.SY↗

arXiv2025

Learning to Drift with Individual Wheel Drive: Maneuvering Autonomous Vehicle at the Handling Limits

Drifting, characterized by controlled vehicle motion at high sideslip angles, is crucial for safely handling emergency scenarios at the friction limits. While recent reinforcement learning approaches show promise for drifting control, they struggle with the significant simulation-to-reality gap, as policies that perform well in simulation often fail when transferred to physical systems. In this paper, we present a reinforcement learning framework with GPU-accelerated parallel simulation and systematic domain randomization that effectively bridges the gap. The proposed approach is validated on both simulation and a custom-designed and open-sourced 1/10 scale Individual Wheel Drive (IWD) RC car platform featuring independent wheel speed control. Experiments across various scenarios from steady-state circular drifting to direction transitions and variable-curvature path following demonstrate that our approach achieves precise trajectory tracking while maintaining controlled sideslip angles throughout complex maneuvers in both simulated and real-world environments.

cs.RO↗

arXiv2025

Constrained Gaussian Process Motion Planning via Stein Variational Newton Inference

Gaussian Process Motion Planning (GPMP) is a widely used framework for generating smooth trajectories within a limited compute time--an essential requirement in many robotic applications. However, traditional GPMP approaches often struggle with enforcing hard nonlinear constraints and rely on Maximum a Posteriori (MAP) solutions that disregard the full Bayesian posterior. This limits planning diversity and ultimately hampers decision-making. Recent efforts to integrate Stein Variational Gradient Descent (SVGD) into motion planning have shown promise in handling complex constraints. Nonetheless, these methods still face persistent challenges, such as difficulties in strictly enforcing constraints and inefficiencies when the probabilistic inference problem is poorly conditioned. To address these issues, we propose a novel constrained Stein Variational Gaussian Process Motion Planning (cSGPMP) framework, incorporating a GPMP prior specifically designed for trajectory optimization under hard constraints. Our approach improves the efficiency of particle-based inference while explicitly handling nonlinear constraints. This advancement significantly broadens the applicability of GPMP to motion planning scenarios demanding robust Bayesian inference, strict constraint adherence, and computational efficiency within a limited time. We validate our method on standard benchmarks, achieving an average success rate of 98.57% across 350 planning tasks, significantly outperforming competitive baselines. This demonstrates the ability of our method to discover and use diverse trajectory modes, enhancing flexibility and adaptability in complex environments, and delivering significant improvements over standard baselines without incurring major computational costs.

cs.RO↗

arXiv2024

Regret Analysis with Almost Sure Convergence for OBF-ARX Filter

This paper considers the output prediction problem for an unknown Linear Time-Invariant (LTI) system. In particular, we focus our attention on the OBF-ARX filter, whose transfer function is a linear combination of Orthogonal Basis Functions (OBFs), with the coefficients determined by solving a least-squares regression. We prove that the OBF-ARX filter is an accurate approximation of the Kalman Filter (KF) by quantifying its online performance. Specifically, we analyze the average regret between the OBF-ARX filter and the KF, proving that the average regret over $N$ time steps converges to the asymptotic bias at the speed of $O(N^{-0.5+ε})$ almost surely for all $ε>0$. Then, we establish an upper bound on the asymptotic bias, demonstrating that it decreases exponentially with the number of OBF bases, and the decreasing rate $τ(\boldsymbolλ, \boldsymbolμ)$ explicitly depends on the poles of both the KF and the OBF. Numerical results on diffusion processes validate the derived bounds.

math.OC↗

arXiv2024

Generalized Activation via Multivariate Projection

Activation functions are essential to introduce nonlinearity into neural networks, with the Rectified Linear Unit (ReLU) often favored for its simplicity and effectiveness. Motivated by the structural similarity between a shallow Feedforward Neural Network (FNN) and a single iteration of the Projected Gradient Descent (PGD) algorithm, a standard approach for solving constrained optimization problems, we consider ReLU as a projection from R onto the nonnegative half-line R+. Building on this interpretation, we extend ReLU by substituting it with a generalized projection operator onto a convex cone, such as the Second-Order Cone (SOC) projection, thereby naturally extending it to a Multivariate Projection Unit (MPU), an activation function with multiple inputs and multiple outputs. We further provide mathematical proof establishing that FNNs activated by SOC projections outperform those utilizing ReLU in terms of expressive power. Experimental evaluations on widely-adopted architectures further corroborate MPU's effectiveness against a broader range of existing activation functions.

cs.LG↗

arXiv2023

Consecutive Inertia Drift of Autonomous RC Car via Primitive-based Planning and Data-driven Control

Inertia drift is an aggressive transitional driving maneuver, which is challenging due to the high nonlinearity of the system and the stringent requirement on control and planning performance. This paper presents a solution for the consecutive inertia drift of an autonomous RC car based on primitive-based planning and data-driven control. The planner generates complex paths via the concatenation of path segments called primitives, and the controller eases the burden on feedback by interpolating between multiple real trajectories with different initial conditions into one near-feasible reference trajectory. The proposed strategy is capable of drifting through various paths containing consecutive turns, which is validated in both simulation and reality.

cs.RO↗

arXiv2023

Linear Model Predictive Control under Continuous Path Constraints via Parallelized Primal-Dual Hybrid Gradient Algorithm

In this paper, we consider a Model Predictive Control (MPC) problem of a continuous-time linear time-invariant system subject to continuous-time path constraints on the states and the inputs. By leveraging the concept of differential flatness, we can replace the differential equations governing the system with linear mapping between the states, inputs, and flat outputs (including their derivatives). The flat outputs are then parameterized by piecewise polynomials, and the model predictive control problem can be equivalently transformed into a Semi-Definite Programming (SDP) problem via Sum-of-Squares (SOS), ensuring constraint satisfaction at every continuous-time interval. We further note that the SDP problem contains a large number of small-size semi-definite matrices as optimization variables. To address this, we develop a Primal-Dual Hybrid Gradient (PDHG) algorithm that can be efficiently parallelized to speed up the optimization procedure. Simulation results on a quadruple-tank process demonstrate that our formulation can guarantee strict constraint satisfaction, while the standard MPC controller based on the discretized system may violate the constraint inside a sampling period. Moreover, the computational speed superiority of our proposed algorithm is collaborated by numerical simulation.

math.OC↗

arXiv2023

Markov Parameter Identification via Chebyshev Approximation

This paper proposes an identification algorithm for Single Input Single Output (SISO) Linear Time-Invariant (LTI) systems. In the noise-free setting, where the first $T$ Markov parameters can be precisely estimated, all Markov parameters can be inferred by the linear combination of the known $T$ Markov parameters, of which the coefficients are obtained by solving the uniform polynomial approximation problem, and the upper bound of the asymptotic identification bias is provided. For the finite-time identification scenario, we cast the system identification problem with noisy Markov parameters into a regularized uniform approximation problem. Numerical results demonstrate that the proposed algorithm outperforms the conventional Ho-Kalman Algorithm for the finite-time identification scenario while the asymptotic bias remains negligible.

math.OC↗

arXiv2021

Black-Box Quantum State Preparation with Inverse Coefficients

Black-box quantum state preparation is a fundamental building block for many higher-level quantum algorithms, which is applied to transduce the data from computational basis into amplitude. Here we present a new algorithm for performing black-box state preparation with inverse coefficients based on the technique of inequality test. This algorithm can be used as a subroutine to perform the controlled rotation stage of the Harrow-Hassidim-Lloyd (HHL) algorithm and the associated matrix inversion algorithms with exceedingly low cost. Furthermore, we extend this approach to address the general black-box state preparation problem where the transduced coefficient is a general non-linear function. The present algorithm greatly relieves the need to do arithmetic and the error is only resulted from the truncated error of binary string. It is expected that our algorithm will find wide usage both in the NISQ and fault-tolerant quantum algorithms.

quant-ph↗

arXiv2020

A Multi-resolution Model for Histopathology Image Classification and Localization with Multiple Instance Learning

Histopathological images provide rich information for disease diagnosis. Large numbers of histopathological images have been digitized into high resolution whole slide images, opening opportunities in developing computational image analysis tools to reduce pathologists' workload and potentially improve inter- and intra- observer agreement. Most previous work on whole slide image analysis has focused on classification or segmentation of small pre-selected regions-of-interest, which requires fine-grained annotation and is non-trivial to extend for large-scale whole slide analysis. In this paper, we proposed a multi-resolution multiple instance learning model that leverages saliency maps to detect suspicious regions for fine-grained grade prediction. Instead of relying on expensive region- or pixel-level annotations, our model can be trained end-to-end with only slide-level labels. The model is developed on a large-scale prostate biopsy dataset containing 20,229 slides from 830 patients. The model achieved 92.7% accuracy, 81.8% Cohen's Kappa for benign, low grade (i.e. Grade group 1) and high grade (i.e. Grade group >= 2) prediction, an area under the receiver operating characteristic curve (AUROC) of 98.2% and an average precision (AP) of 97.4% for differentiating malignant and benign slides. The model obtained an AUROC of 99.4% and an AP of 99.8% for cancer detection on an external dataset.

eess.IV↗