arXiv subjects

Xi Xiong

Publications and source records attributed to Xi Xiong.

At least 19 recordsLinked to original sources

arXiv2026

Offline Multi-Agent Reinforcement Learning with a Physics-Informed World Model for Cooperative Mixed Traffic Control

This study investigates cooperative control of connected and automated vehicles (CAVs) at partially observable highway bottlenecks in mixed traffic, aiming to mitigate congestion without relying on complete global traffic states or online trial-and-error. We propose a physics-informed world model-based offline multi-agent reinforcement learning framework that reconstructs a physically interpretable global traffic state from local CAV observation-action histories, with coupled macroscopic-microscopic traffic dynamics providing physics-based supervision. A probabilistic ensemble world model learns traffic-state transitions and system rewards, while model disagreement quantifies epistemic uncertainty. Multi-step imagined rollouts with pessimistic rewards and uncertainty-driven truncation are then used for offline policy learning. Experiments in a SUMO-based on-ramp bottleneck using approximately $1\times10^6$ offline transitions show that physics supervision improves state reconstruction and world-model prediction accuracy.

cs.MA↗

arXiv2026

Dreamer-SAC: Off-Policy Learning in Latent World Models for Sample-Efficient Autonomous Driving

Sample-efficient reinforcement learning for autonomous driving is often limited by the trade-off between data efficiency and model bias. While world models reduce the reliance on costly environment interactions, policy optimization over learned dynamics remains sensitive to prediction errors. This paper proposes the Dreamer-SAC framework, which integrates a recurrent state-space world model with an off-policy soft actor-critic algorithm trained directly in latent space. The framework uses a combination of real interactions and short-horizon generated trajectories with n-step target estimation and multi-objective supervision. Evaluated in autonomous driving scenarios with objectives encompassing driving efficiency and safety, the proposed framework consistently outperforms representative reinforcement learning baselines, including DreamerV3, SAC, and PPO, while achieving improved performance with substantially fewer real environment interactions. Experiments reveal an inverted-U relationship between rollout horizon and policy performance, where short-horizon latent rollouts achieve the best trade-off between additional training signals and accumulated model bias. Furthermore, n-step target estimation demonstrates more effectiveness over one-step temporal-difference targets in exploiting predicted experience for value learning.

cs.LG↗

arXiv2026

3DCodeBench: Benchmarking Agentic Procedural 3D Modeling Via Code

Procedural 3D modeling through code is emerging as a versatile paradigm, offering deterministic, engine-ready, and precisely editable assets that neural 3D generators inherently lack. Authoring such procedural content, however, demands deep expertise in 3D software APIs, parametric design, and code-level geometric reasoning. In this paper, we propose 3DCodeBench, a systematic benchmark for evaluating vision-language model (VLM) agents for procedural 3D generation in 3D modeling software. Specifically, 3DCodeBench evaluates how effectively 12 advanced VLMs can serve as procedural 3D modelers by translating text and image references into procedural code for 3D modeling software. Recognizing that automated metrics may not fully capture the perceptual quality of 3D shapes, we build 3DCodeArena, a ranking platform based on pairwise human preferences over generated 3D outputs. From extensive evaluations and results, we observe that: (1) Failures mostly arise from API mismatches, while successful renders still suffer from disconnected or floating 3D geometric components. (2) Test-time scaling, such as higher thinking budgets and multi-turn refinement, improves performance overall. Our findings highlight a critical need for high-quality procedural coding data to advance commercial VLMs. Furthermore, effective procedural 3D modeling requires a robust execution environment that provides high-fidelity feedback for iterative refinement. We release 3DCodeBench, including the curated large-scale dataset of multimodal (text/image) prompts, procedural code, 3D object triplets, evaluation protocol, and the public 3DCodeArena platform as a foundational toolkit for exploring VLM-based procedural 3D modelers.

cs.CV↗

arXiv2026

Kinematics-Aware Latent World Models for Data-Efficient Autonomous Driving

Data-efficient learning remains a central challenge in autonomous driving due to the high cost and safety risks of large-scale real-world interaction. Although world-model-based reinforcement learning enables policy optimization through latent imagination, existing approaches often lack explicit mechanisms to encode spatial and kinematic structure essential for driving tasks. In this work, we build upon the Recurrent State-Space Model (RSSM) and propose a kinematics-aware latent world model framework for autonomous driving. Vehicle kinematic information is incorporated into the observation encoder to ground latent transitions in physically meaningful motion dynamics, while geometry-aware supervision regularizes the RSSM latent state to capture task-relevant spatial structure beyond pixel reconstruction. The resulting structured latent dynamics improve long-horizon imagination fidelity and stabilize policy optimization. Experiments in a driving simulation benchmark demonstrate consistent gains over both model-free and pixel-based world-model baselines in terms of sample efficiency and driving performance. Ablation studies further verify that the proposed design enhances spatial representation quality within the latent space. These results suggest that integrating kinematic grounding into RSSM-based world models provides a scalable and physically grounded paradigm for autonomous driving policy learning.

cs.RO↗

arXiv2025

Probe-and-Release Coordination of Platoons at Highway Bottlenecks with Unknown Parameters

This paper considers coordination of platoons of connected and autonomous vehicles (CAVs) at mixed-autonomy bottlenecks in the face of three practically important factors, viz. time-varying traffic demand, random CAV platoon sizes, and capacity breakdowns. Platoon coordination is essential to smoothen the interaction between CAV platoons and non-CAV traffic. Based on a fluid queuing model, we develop a "probe-and-release" algorithm that simultaneously estimates environmental parameters and coordinates CAV platoons for traffic stabilization. We show that this algorithm ensures bounded estimation errors and bounded traffic queues. The proof builds on a Lyapunov function that jointly penalizes estimation errors and traffic queues and a drift argument for an embedded Markov process. We validate the proposed algorithm in a standard micro-simulation environment and compare against a representative deep reinforcement learning method in terms of control performance and computational efficiency.

eess.SY↗

arXiv2025

Optimizing Highway Traffic Flow in Mixed Autonomy: A Multiagent Truncated Rollout Approach

The development of connected and autonomous vehicles (CAVs) offers substantial opportunities to enhance traffic efficiency. However, in mixed autonomy environments where CAVs coexist with human-driven vehicles (HDVs), achieving efficient coordination among CAVs remains challenging due to heterogeneous driving behaviors. To address this, this paper proposes a multiagent truncated rollout approach that enhances CAV speed coordination to improve highway throughput while reducing computational overhead. In this approach, a traffic density evolution equation is formulated that comprehensively accounts for the presence or absence of CAVs, and a distributed coordination control framework is established accordingly. By incorporating kinematic information from neighbor agents and employing an agent-by-agent sequential solution mechanism, our method enables explicit cooperation among CAVs. Furthermore, we introduce a truncated rollout scheme that adaptively shortens the optimization horizon based on the evaluation of control sequences. This significantly reduces the time complexity, thereby improving real-time performance and scalability. Theoretical analysis provides rigorous guarantees on the stability and performance improvement of the system. Simulations conducted on real-world bottleneck scenarios demonstrate that, in large-scale mixed traffic flows, the proposed method outperforms conventional model predictive control methods by reducing both the average travel time in the bottleneck area and overall computational time, highlighting its strong potential for practical deployment.

cs.MA↗

arXiv2025

Multi-armed Bandit for Stochastic Shortest Path in Mixed Autonomy

In mixed-autonomy traffic networks, autonomous vehicles (AVs) are required to make sequential routing decisions under uncertainty caused by dynamic and heterogeneous interactions with human-driven vehicles (HDVs). Early-stage greedy decisions made by AVs during interactions with the environment often result in insufficient exploration, leading to failures in discovering globally optimal strategies. The exploration-exploitation balancing mechanism inherent in multi-armed bandit (MAB) methods is well-suited for addressing such problems. Based on the Real-Time Dynamic Programming (RTDP) framework, we introduce the Upper Confidence Bound (UCB) exploration strategy from the MAB paradigm and propose a novel algorithm. We establish the path-level regret upper bound under the RTDP framework, which guarantees the worst-case convergence of the proposed algorithm. Extensive numerical experiments conducted on a real-world local road network in Shanghai demonstrate that the proposed algorithm effectively overcomes the failure of standard RTDP to converge to the optimal policy under highly stochastic environments. Moreover, compared to the standard Value Iteration (VI) framework, the RTDP-based framework demonstrates superior computational efficiency. Our results highlight the effectiveness of the proposed algorithm in routing within large-scale stochastic mixed-autonomy environments.

math.OC↗

arXiv2025

A Large Language Model-Enhanced Q-learning for Capacitated Vehicle Routing Problem with Time Windows

The Capacitated Vehicle Routing Problem with Time Windows (CVRPTW) is a classic NP-hard combinatorial optimization problem widely applied in logistics distribution and transportation management. Its complexity stems from the constraints of vehicle capacity and time windows, which pose significant challenges to traditional approaches. Advances in Large Language Models (LLMs) provide new possibilities for finding approximate solutions to CVRPTW. This paper proposes a novel LLM-enhanced Q-learning framework to address the CVRPTW with real-time emergency constraints. Our solution introduces an adaptive two-phase training mechanism that transitions from the LLM-guided exploration phase to the autonomous optimization phase of Q-network. To ensure reliability, we design a three-tier self-correction mechanism based on the Chain-of-Thought (CoT) for LLMs: syntactic validation, semantic verification, and physical constraint enforcement. In addition, we also prioritized replay of the experience generated by LLMs to amplify the regulatory role of LLMs in the architecture. Experimental results demonstrate that our framework achieves a 7.3\% average reduction in cost compared to traditional Q-learning, with fewer training steps required for convergence.

cs.LG↗

arXiv2024

Reinforcement Learning from Human Feedback for Lane Changing of Autonomous Vehicles in Mixed Traffic

The burgeoning field of autonomous driving necessitates the seamless integration of autonomous vehicles (AVs) with human-driven vehicles, calling for more predictable AV behavior and enhanced interaction with human drivers. Human-like driving, particularly during lane-changing maneuvers on highways, is a critical area of research due to its significant impact on safety and traffic flow. Traditional rule-based decision-making approaches often fail to encapsulate the nuanced boundaries of human behavior in diverse driving scenarios, while crafting reward functions for learning-based methods introduces its own set of complexities. This study investigates the application of Reinforcement Learning from Human Feedback (RLHF) to emulate human-like lane-changing decisions in AVs. An initial RL policy is pre-trained to ensure safe lane changes. Subsequently, this policy is employed to gather data, which is then annotated by humans to train a reward model that discerns lane changes aligning with human preferences. This human-informed reward model supersedes the original, guiding the refinement of the policy to reflect human-like preferences. The effectiveness of RLHF in producing human-like lane changes is demonstrated through the development and evaluation of conservative and aggressive lane-changing models within obstacle-rich environments and mixed autonomy traffic scenarios. The experimental outcomes underscore the potential of RLHF to diversify lane-changing behaviors in AVs, suggesting its viability for enhancing the integration of AVs into the fabric of human-driven traffic.

cs.CE↗

arXiv2024

PaliGemma: A versatile 3B VLM for transfer

PaliGemma is an open Vision-Language Model (VLM) that is based on the SigLIP-So400m vision encoder and the Gemma-2B language model. It is trained to be a versatile and broadly knowledgeable base model that is effective to transfer. It achieves strong performance on a wide variety of open-world tasks. We evaluate PaliGemma on almost 40 diverse tasks including standard VLM benchmarks, but also more specialized tasks such as remote-sensing and segmentation.

cs.CV↗

arXiv2024

iLLM-TSC: Integration reinforcement learning and large language model for traffic signal control policy improvement

Urban congestion remains a critical challenge, with traffic signal control (TSC) emerging as a potent solution. TSC is often modeled as a Markov Decision Process problem and then solved using reinforcement learning (RL), which has proven effective. However, the existing RL-based TSC system often overlooks imperfect observations caused by degraded communication, such as packet loss, delays, and noise, as well as rare real-life events not included in the reward function, such as unconsidered emergency vehicles. To address these limitations, we introduce a novel integration framework that combines a large language model (LLM) with RL. This framework is designed to manage overlooked elements in the reward function and gaps in state information, thereby enhancing the policies of RL agents. In our approach, RL initially makes decisions based on observed data. Subsequently, LLMs evaluate these decisions to verify their reasonableness. If a decision is found to be unreasonable, it is adjusted accordingly. Additionally, this integration approach can be seamlessly integrated with existing RL-based TSC systems without necessitating modifications. Extensive testing confirms that our approach reduces the average waiting time by $17.5\%$ in degraded communication conditions as compared to traditional RL methods, underscoring its potential to advance practical RL applications in intelligent transportation systems. The related code can be found at \url{https://github.com/Traffic-Alpha/iLLM-TSC}.

cs.AI↗

arXiv2024

Traffic Signal Cycle Control with Centralized Critic and Decentralized Actors under Varying Intervention Frequencies

Traffic congestion in urban areas is a significant problem, leading to prolonged travel times, reduced efficiency, and increased environmental concerns. Effective traffic signal control (TSC) is a key strategy for reducing congestion. Unlike most TSC systems that rely on high-frequency control, this study introduces an innovative joint phase traffic signal cycle control method that operates effectively with varying control intervals. Our method features an adjust all phases action design, enabling simultaneous phase changes within the signal cycle, which fosters both immediate stability and sustained TSC effectiveness, especially at lower frequencies. The approach also integrates decentralized actors to handle the complexity of the action space, with a centralized critic to ensure coordinated phase adjusting. Extensive testing on both synthetic and real-world data across different intersection types and signal setups shows that our method significantly outperforms other popular techniques, particularly at high control intervals. Case studies of policies derived from traffic data further illustrate the robustness and reliability of our proposed method.

eess.SY↗

arXiv2024

A Multi-Agent Rollout Approach for Highway Bottleneck Decongestion in Mixed Autonomy

The integration of autonomous vehicles (AVs) into the existing transportation infrastructure offers a promising solution to alleviate congestion and enhance mobility. This research explores a novel approach to traffic optimization by employing a multi-agent rollout approach within a mixed autonomy environment. The study concentrates on coordinating the speed of human-driven vehicles by longitudinally controlling AVs, aiming to dynamically optimize traffic flow and alleviate congestion at highway bottlenecks in real-time. We model the problem as a decentralized partially observable Markov decision process (Dec-POMDP) and propose an improved multi-agent rollout algorithm. By employing agent-by-agent policy iterations, our approach implicitly considers cooperation among multiple agents and seamlessly adapts to complex scenarios where the number of agents dynamically varies. Validated in a real-world network with varying AV penetration rates and traffic flow, the simulations demonstrate that the multi-agent rollout algorithm significantly enhances performance, reducing average travel time on bottleneck segments by 9.42% with a 10% AV penetration rate.

cs.MA↗

arXiv2024

Dyna-Style Learning with A Macroscopic Model for Vehicle Platooning in Mixed-Autonomy Traffic

Platooning of connected and autonomous vehicles (CAVs) plays a vital role in modernizing highways, ushering in enhanced efficiency and safety. This paper explores the significance of platooning in smart highways, employing a coupled partial differential equation (PDE) and ordinary differential equation (ODE) model to elucidate the complex interaction between bulk traffic flow and CAV platoons. Our study focuses on developing a Dyna-style planning and learning framework tailored for platoon control, with a specific goal of reducing fuel consumption. By harnessing the coupled PDE-ODE model, we improve data efficiency in Dyna-style learning through virtual experiences. Simulation results validate the effectiveness of our macroscopic model in modeling platoons within mixed-autonomy settings, demonstrating a notable $10.11\%$ reduction in vehicular fuel consumption compared to conventional approaches.

cs.LG↗

arXiv2023

UniTSA: A Universal Reinforcement Learning Framework for V2X Traffic Signal Control

Traffic congestion is a persistent problem in urban areas, which calls for the development of effective traffic signal control (TSC) systems. While existing Reinforcement Learning (RL)-based methods have shown promising performance in optimizing TSC, it is challenging to generalize these methods across intersections of different structures. In this work, a universal RL-based TSC framework is proposed for Vehicle-to-Everything (V2X) environments. The proposed framework introduces a novel agent design that incorporates a junction matrix to characterize intersection states, making the proposed model applicable to diverse intersections. To equip the proposed RL-based framework with enhanced capability of handling various intersection structures, novel traffic state augmentation methods are tailor-made for signal light control systems. Finally, extensive experimental results derived from multiple intersection configurations confirm the effectiveness of the proposed framework. The source code in this work is available at https://github.com/wmn7/Universal_Light

eess.SY↗

arXiv2023

Combining Policy Gradient and Safety-Based Control for Autonomous Driving

With the advancement of data-driven techniques, addressing continuous con-trol challenges has become more efficient. However, the reliance of these methods on historical data introduces the potential for unexpected decisions in novel scenarios. To enhance performance in autonomous driving and collision avoidance, we propose a symbiotic fusion of policy gradient with safety-based control. In this study, we em-ploy the Deep Deterministic Policy Gradient (DDPG) algorithm to enable autono-mous driving in the absence of surrounding vehicles. By training the vehicle's driving policy within a stable and familiar environment, a robust and efficient learning pro-cess is achieved. Subsequently, an artificial potential field approach is utilized to formulate a collision avoidance algorithm, accounting for the presence of surround-ing vehicles. Furthermore, meticulous consideration is given to path tracking meth-ods. The amalgamation of these approaches demonstrates substantial performance across diverse scenarios, underscoring its potential for advancing autonomous driving while upholding safety standards.

cs.RO↗

arXiv2023

PaLI-3 Vision Language Models: Smaller, Faster, Stronger

This paper presents PaLI-3, a smaller, faster, and stronger vision language model (VLM) that compares favorably to similar models that are 10x larger. As part of arriving at this strong performance, we compare Vision Transformer (ViT) models pretrained using classification objectives to contrastively (SigLIP) pretrained ones. We find that, while slightly underperforming on standard image classification benchmarks, SigLIP-based PaLI shows superior performance across various multimodal benchmarks, especially on localization and visually-situated text understanding. We scale the SigLIP image encoder up to 2 billion parameters, and achieves a new state-of-the-art on multilingual cross-modal retrieval. We hope that PaLI-3, at only 5B parameters, rekindles research on fundamental pieces of complex VLMs, and could fuel a new generation of scaled-up models.

cs.CV↗

arXiv2023

An Approximate Dynamic Programming Approach to Vehicle Platooning Coordination in Networks

Platooning connected and autonomous vehicles (CAVs) provide significant benefits in terms of traffic efficiency and fuel economy. However, most existing platooning systems assume the availability of pre-determined plans, which is not feasible in real-time scenarios. In this paper, we address this issue in time-dependent networks by formulating a Markov decision process at each junction, aiming to minimize travel time and fuel consumption. Initially, we analyze coordinated platooning without routing to explore the cooperation among controllers on an identical path. We propose two novel approaches based on approximate dynamic programming, offering suboptimal control in the context of a stochastic finite horizon problem. The results demonstrate the superiority of the approximation in the policy space. Furthermore, we investigate platooning in a network setting, where speed profiles and routes are determined simultaneously. To simplify the problem, we decouple the action space by prioritizing routing decisions based on travel time estimation. We subsequently employ the aforementioned policy approximation to determine speed profiles, considering essential parameters such as travel times. Our simulation results in SUMO indicate that our method yields better performance than conventional approaches, leading to potential travel cost savings of up to 40%. Additionally, we evaluate the resilience of our approach in dynamically changing networks, affirming its ability to maintain efficient platooning operations.

eess.SY↗