arXiv subjects

Yiyuan Wang

Publications and source records attributed to Yiyuan Wang.

At least 19 recordsLinked to original sources

arXiv2026

A Nested Iterative Algorithm for Zero-Sum Linear-Quadratic Stochastic Differential Games in an Infinite Horizon

This paper proposes a new algorithm to compute closed-loop saddle points for infinite-horizon zero-sum linear-quadratic stochastic differential games via structural decoupling. Specifically, we develop a nested iterative scheme that generates a monotonically increasing matrix sequence to decompose the original problem into coupled subproblems. By sequentially solving the stabilizing solutions of associated algebraic Riccati equations for each subproblem, we recover the original problem's stabilizing solution and rigorously prove sequence convergence. A numerical example further validates the effectiveness of the proposed method. To the best of our knowledge, this work extends the classical setting and provides the first general-purpose computational approach for this class of problems.

math.OC↗

arXiv2026

ReCal3R: Reliability-Calibrated Learning Rates for Streaming 3D Reconstruction

Streaming 3D reconstruction relies on a compact recurrent scene state to process long image streams in linear time and bounded memory. However, repeated updates can gradually corrupt this state, causing reliable historical information to be overwritten by noisy or ambiguous observations. We introduce ReCal3R, a reliability-calibrated learning rate method for recurrent 3D reconstruction. Instead of directly applying a candidate learning rate, our method estimates state token reliability from the maintained scene state and uses it to calibrate a candidate learning rate derived from token alignment, state reconstruction residual, and recent update pressure. The resulting token-wise learning rate interpolates between a conservative base rate and the candidate rate, suppressing aggressive updates on unreliable tokens while preserving adaptation to informative frames. Applied to CUT3R as a training-free calibration rule, ReCal3R reaches strong performance on long sequences in pose, depth, and reconstruction quality, including a 3.7$\times$ reduction in ATE, with comparable runtime and memory. Code is available at: https://github.com/Powertony102/ReCal3R.

cs.CV↗

arXiv2026

An Iterative Computational Framework for Infinite-Horizon Mean-Field Linear-Quadratic Zero-Sum Stochastic Differential Games

This work develops an iterative computational framework to obtain saddle-point solutions for infinite-horizon two-person mean-field linear-quadratic zero-sum stochastic differential games. By generalizing classical iterative framework, we construct a monotonically increasing matrix sequence to decouple the strongly coupled, analytically intractable original problem into a set of tractable subproblems. By sequentially computing the stabilizing solutions of the coupled algebraic Riccati equations for these subproblems, we can further derive the solution to the original problem. Rigorous convergence analysis is established to validate the proposed iterative algorithm. Different from existing algorithms limited to specific simplified setups, this framework proposes the first universal computational paradigm applicable to a broad class of game-driven Riccati equations.

math.OC↗

arXiv2026

QuadBox: Accelerating 3D Gaussian Splatting with Geometry-Aware Boxes

3D Gaussian Splatting (3DGS) has emerged as an advanced technique for real-time novel view synthesis by representing scene geometry and appearance using differentiable Gaussian primitives. However, efficiently computing precise Gaussian-tile intersections remains a critical task in the rasterization pipeline. To this end, we propose QuadBox, a method that leverages four axis-aligned bounding boxes to tightly encapsulate projected Gaussians in a discrete manner. First, we derive a geometry-aware stretching factor that enables the construction of a tile-aligned QuadBox, which covers the elliptical projection and largely excludes irrelevant tiles. Second, we introduce QPass, a single-pass tile traversal algorithm that exhaustively exploits the discrete nature of QuadBox, ensuring that the tile intersection check is performed with simple interval tests. Experiments on public datasets show that our method accelerates the rendering speed of 3DGS by 1.85$\times$. Code is available at \href{https://github.com/Powertony102/QuadBox}{https://github.com/Powertony102/QuadBox}.

cs.CV↗

arXiv2026

From Passersby to Placemaking: Designing Autonomous Vehicle-Pedestrian Encounters for an Urban Shared Space

Autonomous vehicles (AVs) tend to disrupt the atmosphere and pedestrian experience in urban shared spaces, undermining the focus of these spaces on people and placemaking. We investigate how external human-machine interfaces (eHMIs) supporting AV-pedestrian interaction can be extended to consider the characteristics of an urban shared space. Inspired by urban HCI, we devised three place-based eHMI designs that (i) enhance a conventional intent eHMI and (ii) exhibit content and physical integration with the space. In an evaluation study, 25 participants experienced the eHMIs in an immersive simulation of the space via virtual reality and shared their impressions through think-aloud, interviews, and questionnaires. Results showed that the place-based eHMIs had a notable effect on influencing the perception of AV interaction, including aspects like visual aesthetics and sense of reassurance, and on fostering a sense of place, such as social interactivity and the intentionality to coexist. In measuring qualities of pedestrian experience, we found that perceived safety significantly correlated with user experience and affect, including the attractiveness of eHMIs and feelings of pleasantness. The paper opens the avenue for exploring how eHMIs may contribute to the placemaking goals of pedestrian-centric spaces and improve the experience of people encountering AVs within these environments.

cs.HC↗

arXiv2026

S-VGGT: Structure-Aware Subscene Decomposition for Scalable 3D Foundation Models

Feed-forward 3D foundation models face a key challenge: the quadratic computational cost introduced by global attention, which severely limits scalability as input length increases. Concurrent acceleration methods, such as token merging, operate at the token level. While they offer local savings, the required nearest-neighbor searches introduce undesirable overhead. Consequently, these techniques fail to tackle the fundamental issue of structural redundancy dominant in dense capture data. In this work, we introduce \textbf{S-VGGT}, a novel approach that addresses redundancy at the structural frame level, drastically shifting the optimization focus. We first leverage the initial features to build a dense scene graph, which characterizes structural scene redundancy and guides the subsequent scene partitioning. Using this graph, we softly assign frames to a small number of subscenes, guaranteeing balanced groups and smooth geometric transitions. The core innovation lies in designing the subscenes to share a common reference frame, establishing a parallel geometric bridge that enables independent and highly efficient processing without explicit geometric alignment. This structural reorganization provides strong intrinsic acceleration by cutting the global attention cost at its source. Crucially, S-VGGT is entirely orthogonal to token-level acceleration methods, allowing the two to be seamlessly combined for compounded speedups without compromising reconstruction fidelity. Code is available at https://github.com/Powertony102/S-VGGT.

cs.CV↗

arXiv2026

Conversational AI-Enhanced Exploration System to Query Large-Scale Digitised Collections of Natural History Museums

Recent digitisation efforts in natural history museums have produced large volumes of collection data, yet their scale and scientific complexity often hinder public access and understanding. Conventional data management tools, such as databases, restrict exploration through keyword-based search or require specialised schema knowledge. This paper presents a system design that uses conversational AI to query nearly 1.7 million digitised specimen records from the life-science collections of the Australian Museum. Designed and developed through a human-centred design process, the system contains an interactive map for visual-spatial exploration and a natural-language conversational agent that retrieves detailed specimen data and answers collection-specific questions. The system leverages function-calling capabilities of contemporary large language models to dynamically retrieve structured data from external APIs, enabling fast, real-time interaction with extensive yet frequently updated datasets. Our work provides a new approach of connecting large museum collections with natural language-based queries and informs future designs of scientific AI agents for natural history museums.

cs.HC↗

arXiv2026

Reinforcement Learning Method for Zero-Sum Linear-Quadratic Stochastic Differential Games in Infinite Horizons

In this work, we propose, for the first time, a reinforcement learning framework specifically designed for zero-sum linear-quadratic stochastic differential games. This approach offers a generalized solution for scenarios in which accurate system parameters are difficult to obtain, thereby overcoming a key limitation of traditional iterative methods that rely on complete system information. In correspondence with the game-theoretic algebraic Riccati equations associated with the problem, we develop both semi-model-based and model-free reinforcement learning algorithms by combining an iterative solution scheme with dynamic programming principles. Notably, under appropriate rank conditions on data sampling, the convergence of the proposed algorithms is rigorously established through theoretical analysis. Finally, numerical simulations are conducted to verify the effectiveness and feasibility of the proposed method.

math.OC↗

arXiv2026

Designing Drone Interfaces to Assist Pedestrians Crossing Non-Signalised Roads

Recent research highlights the potential of drones to enhance pedestrian experiences, such as aiding navigation and supporting street-level activities. This paper explores the design of drone interfaces to assist pedestrians crossing dangerous roads without designated crosswalks or traffic lights, leveraging drones' ability to monitor and analyse real-time traffic data. Inspired by existing traffic signal systems, the interface communicates safety information through permissive alerts, prohibitive warnings, directional warnings, and collision emergency warnings. These safety cues were integrated into drone interfaces using in-situ projections and drone-equipped screens through an iterative design process. A mixed-methods, within-subjects VR evaluation (n=18) revealed that drone-assisted systems significantly improved pedestrian safety experiences and reduced mental workload compared to a baseline without any crossing aid, with projections outperforming screens. The findings suggest the potential for drone interfaces to be integrated into connected traffic systems. We also offer design recommendations for developing drone interfaces that support safe pedestrian crossings.

cs.HC↗

arXiv2025

A Convergent Algorithm Based on Deterministic Approximation for a Large Class of Regime-Switching Generalized Stochastic Game-Theoretic Riccati Differential Equations

This paper proposes a novel iterative algorithm to compute the stabilizing solution of regime-switching stochastic game-theoretic Riccati differential equations with periodic coefficients. The method decomposes the original complex stochastic problem into a sequence of deterministic subproblems. By sequentially solving for the minimal solutions of the Riccati differential equations in each subproblem, a sequence of matrix-valued functions is constructed. Leveraging the comparison theorem, the monotonicity, boundedness, and convergence of the iterative sequence are rigorously proven. Numerical experiments verifies algorithm effectiveness and stability. To the best of our knowledge, this is the first general computational approach developed for this class of problems.

math.NA↗

arXiv2025

A New Algorithm for Computing the Stabilizing Solution of General Periodic Time-Varying Stochastic Game-Theoretic Riccati Differential Equations

We propose a new algorithm for a broad class of periodic time-varying Stochastic Game-Theoretic Riccati Differential Equations arising in Zero-Sum Linear-Quadratic Stochastic Differential Games. The algorithm is constructed via dual-layer matrix-valued functions iteration sequences, which reformulate the original problem into a set of interconnected bilevel subproblems. By sequentially computing the maximal periodic solutions to the Riccati differential equations associated with each subproblem, we derive the stabilizing periodic solutions for the original problem and rigorously prove the algorithm's convergence. Numerical experiments verifies algorithm effectiveness and stability. This study provides a unified numerical framework for solving a wider range of periodic time-varying Stochastic Game-Theoretic Riccati Differential Equations.

math.NA↗

arXiv2025

Resonant helical dichroism in twisted dielectric metastructures

Circular dichroism, arising from interactions with light fields of opposite spin angular momentum, has become a fundamental tool for molecular characterization. Meanwhile, helical dichroism (HD) - the dichroic response to vortex beams carrying opposite orbital angular momentum (OAM) - offers an alternative approach for probing chiral molecules and photonic structures. Previous demonstrations of HD have been limited to non-resonant light-matter interactions with chiral micro- and nanostructures, leaving the realization of resonance helical dichroism largely unexplored. Here, we present the design and implementation of twisted dielectric metastructures, composed of an array of rotated silicon trimer nanostructures harnessing nonlocal photonic modes with a high quality factor of several dozen that enable strong resonant HD for OAM values up to $10$. We experimentally demonstrate resonantly enhanced HD for strongly focused OAM beams with the magnitude of topological charges from $1$ to $3$. Our findings pave the way for resonant nanophotonics involving OAM beams, unlocking the full potential of structured light for applications in molecular sensing, optical imaging, nonlinear optics, and optical data storage.

physics.optics↗

arXiv2025

Peek into the `White-Box': A Field Study on Bystander Engagement with Urban Robot Uncertainty

Uncertainty inherently exists in the autonomous decision-making process of robots. Involving humans in resolving this uncertainty not only helps robots mitigate it but is also crucial for improving human-robot interactions. However, in public urban spaces filled with unpredictability, robots often face heightened uncertainty without direct human collaborators. This study investigates how robots can engage bystanders for assistance in public spaces when encountering uncertainty and examines how these interactions impact bystanders' perceptions and attitudes towards robots. We designed and tested a speculative `peephole' concept that engages bystanders in resolving urban robot uncertainty. Our design is guided by considerations of non-intrusiveness and eliciting initiative in an implicit manner, considering bystanders' unique role as non-obligated participants in relation to urban robots. Drawing from field study findings, we highlight the potential of involving bystanders to mitigate urban robots' technological imperfections to both address operational challenges and foster public acceptance of urban robots. Furthermore, we offer design implications to encourage bystanders' involvement in mitigating the imperfections.

cs.RO↗

arXiv2024

Robots in the Wild: Contextually-Adaptive Human-Robot Interactions in Urban Public Environments

The increasing transition of human-robot interaction (HRI) context from controlled settings to dynamic, real-world public environments calls for enhanced adaptability in robotic systems. This can go beyond algorithmic navigation or traditional HRI strategies in structured settings, requiring the ability to navigate complex public urban systems containing multifaceted dynamics and various socio-technical needs. Therefore, our proposed workshop seeks to extend the boundaries of adaptive HRI research beyond predictable, semi-structured contexts and highlight opportunities for adaptable robot interactions in urban public environments. This half-day workshop aims to explore design opportunities and challenges in creating contextually-adaptive HRI within these spaces and establish a network of interested parties within the OzCHI research community. By fostering ongoing discussions, sharing of insights, and collaborations, we aim to catalyse future research that empowers robots to navigate the inherent uncertainties and complexities of real-world public interactions.

cs.RO↗

arXiv2024

Exploring the Impact of Interconnected External Interfaces in Autonomous Vehicleson Pedestrian Safety and Experience

Policymakers advocate for the use of external Human-Machine Interfaces (eHMIs) to allow autonomous vehicles (AVs) to communicate their intentions or status. Nonetheless, scalability concerns in complex traffic scenarios arise, such as potentially increasing pedestrian cognitive load or conveying contradictory signals. Building upon precursory works, our study explores 'interconnected eHMIs,' where multiple AV interfaces are interconnected to provide pedestrians with clear and unified information. In a virtual reality study (N=32), we assessed the effectiveness of this concept in improving pedestrian safety and their crossing experience. We compared these results against two conditions: no eHMIs and unconnected eHMIs. Results indicated interconnected eHMIs enhanced safety feelings and encouraged cautious crossings. However, certain design elements, such as the use of the colour red, led to confusion and discomfort. Prior knowledge slightly influenced perceptions of interconnected eHMIs, underscoring the need for refined user education. We conclude with practical implications and future eHMI design research directions.

cs.HC↗

arXiv2024

Designing Wearable Augmented Reality Concepts to Support Scalability in Autonomous Vehicle-Pedestrian Interaction

Wearable augmented reality (AR) offers new ways for supporting the interaction between autonomous vehicles (AVs) and pedestrians due to its ability to integrate timely and contextually relevant data into the user's field of view. This article presents novel wearable AR concepts that assist crossing pedestrians in multi-vehicle scenarios where several AVs frequent the road from both directions. Three concepts with different communication approaches for signaling responses from multiple AVs to a crossing request, as well as a conventional pedestrian push button, were simulated and tested within a virtual reality environment. The results showed that wearable AR is a promising way to reduce crossing pedestrians' cognitive load when the design offers both individual AV responses and a clear signal to cross. The willingness of pedestrians to adopt a wearable AR solution, however, is subject to different factors, including costs, data privacy, technical defects, liability risks, maintenance duties, and form factors. We further found that all participants favored sending a crossing request to AVs rather than waiting for the vehicles to detect their intentions-pointing to an important gap and opportunity in the current AV-pedestrian interaction literature.

cs.HC↗

arXiv2024

Pedestrian-Vehicle Interaction in Shared Space: Insights for Autonomous Vehicles

Shared space reduces segregation between vehicles and pedestrians and encourages them to share roads without imposed traffic rules. The behaviour of road users (RUs) is then controlled by social norms, and interactions are more versatile than on traditional roads. Autonomous vehicles (AVs) will need to adapt to these norms to become socially acceptable RUs in shared spaces. However, to date, there is not much research into pedestrian-vehicle interaction in shared-space environments, and prior efforts have predominantly focused on traditional roads and crossing scenarios. We present a video observation investigating pedestrian reactions to a small, automation-capable vehicle driven manually in shared spaces based on a long-term naturalistic driving dataset. We report various pedestrian reactions (from movement adjustment to prosocial behaviour) and situations pertinent to shared spaces at this early stage. Insights drawn can serve as a foundation to support future AVs navigating shared spaces, especially those with a high pedestrian focus.

cs.HC↗

arXiv2024

How Can Autonomous Vehicles Convey Emotions to Pedestrians? A Review of Emotionally Expressive Non-Humanoid Robots

In recent years, researchers and manufacturers have started to investigate ways to enable autonomous vehicles (AVs) to interact with nearby pedestrians in compensation for the absence of human drivers. The majority of these efforts focuses on external human-machine interfaces (eHMIs), using different modalities, such as light patterns or on-road projections, to communicate the AV's intent and awareness. In this paper, we investigate the potential role of affective interfaces to convey emotions via eHMIs. To date, little is known about the role that affective interfaces can play in supporting AV-pedestrian interaction. However, emotions have been employed in many smaller social robots, from domestic companions to outdoor aerial robots in the form of drones. To develop a foundation for affective AV-pedestrian interfaces, we reviewed the emotional expressions of non-humanoid robots in 25 articles published between 2011 and 2021. Based on findings from the review, we present a set of considerations for designing affective AV-pedestrian interfaces and highlight avenues for investigating these opportunities in future studies.

cs.HC↗