Guangyao Shi

Guangyao Shi, Pratap Tokekar

We study the problem of learning a function that maps context observations (input) to parameters of a submodular function (output). Our motivating case study is a specific type of vehicle routing problem, in which a team of Unmanned Ground Vehicles (UGVs) can serve as mobile charging stations to recharge a team of Unmanned Ground Vehicles (UAVs) that execute persistent monitoring tasks. {We want to learn the mapping from observations of UAV task routes and wind field to the parameters of a submodular objective function, which describes the distribution of landing positions of the UAVs .} Traditionally, such a learning problem is solved independently as a prediction phase without considering the downstream task optimization phase. However, the loss function used in prediction may be misaligned with our final goal, i.e., a good routing decision. Good performance in the isolated prediction phase does not necessarily lead to good decisions in the downstream routing task. In this paper, we propose a framework that incorporates task optimization as a differentiable layer in the prediction phase. Our framework allows end-to-end training of the prediction model without using engineered intermediate loss that is targeted only at the prediction performance. In the proposed framework, task optimization (submodular maximization) is made differentiable by introducing stochastic perturbations into deterministic algorithms (i.e., stochastic smoothing). We demonstrate the efficacy of the proposed framework using synthetic data. Experimental results of the mobile charging station routing problem show that the proposed framework can result in better routing decisions, e.g. the average number of UAVs recharged increases, compared to the prediction-optimization separate approach.

Xiyang Wu, Guangyao Shi, Qingzi Wang et al.

Vision-language-action (VLA) models enable robots to follow natural-language instructions grounded in visual observations, but the instruction channel also introduces a critical vulnerability: small textual perturbations can alter downstream robot behavior. Systematic robustness evaluation therefore requires a black-box attacker that can generate minimal yet effective instruction edits across diverse VLA models. To this end, we present SABER, an agent-centric approach for automatically generating instruction-based adversarial attacks on VLA models under bounded edit budgets. SABER uses a GRPO-trained ReAct attacker to generate small, plausible adversarial instruction edits using character-, token-, and prompt-level tools under a bounded edit budget that induces targeted behavioral degradation, including task failure, unnecessarily long execution, and increased constraint violations. On the LIBERO benchmark across six state-of-the-art VLA models, SABER reduces task success by 20.6%, increases action-sequence length by 55%, and raises constraint violations by 33%, while requiring 21.1% fewer tool calls and 54.7% fewer character edits than strong GPT-based baselines. These results show that small, plausible instruction edits are sufficient to substantially degrade robot execution, and that an agentic black-box pipeline offers a practical, scalable, and adaptive approach for red-teaming robotic foundation models.

Zongxia Li, Xiyang Wu, Hongyang Du et al.

Multimodal Vision Language Models (VLMs) have emerged as a transformative topic at the intersection of computer vision and natural language processing, enabling machines to perceive and reason about the world through both visual and textual modalities. For example, models such as CLIP, Claude, and GPT-4V demonstrate strong reasoning and understanding abilities on visual and textual data and beat classical single modality vision models on zero-shot classification [93]. With their rapid advancements in research and growing popularity in various applications, we provide a comprehensive survey of VLMs. Specifically, we provide a systematic overview of VLMs in the following aspects: [1] model information of the major VLMs developed up to 2025; [2] the transition of VLM architectures and the newest VLM alignment methods; [3] summary and categorization of the popular benchmarks and evaluation metrics of VLMs; [4] the challenges and issues faced by current VLMs such as hallucination, alignment, fairness, and safety. Detailed collections including papers and model repository links are listed in https://github.com/zli12321/Vision-Language-Models-Overview.

Xiyang Wu, Zongxia Li, Guangyao Shi et al.

Long horizon interactive environments are a testbed for evaluating agents skill usage abilities. These environments demand multi step reasoning, the chaining of multiple skills over many timesteps, and robust decision making under delayed rewards and partial observability. Games are a good testbed for evaluating agent skill usage in environments. Large Language Models (LLMs) offer a promising alternative as game playing agents, but they often struggle with consistent long horizon decision making because they lack a mechanism to discover, retain, and reuse structured skills across episodes. We present COSPLAY, a co evolution framework in which an LLM decision agent retrieves skills from a learnable skill bank to guide action taking, while an agent managed skill pipeline discovers reusable skills from the agents unlabeled rollouts to form a skill bank. Our framework improves both the decision agent to learn better skill retrieval and action generation, while the skill bank agent continually extracts, refines, and updates skills together with their contracts. Experiments across six game environments show that COSPLAY with an 8B base model achieves over 25.1 percent average reward improvement against four frontier LLM baselines on single player game benchmarks while remaining competitive on multi player social reasoning games.

Peng Chen, Wenxuan He, Feng Qian et al.

In the hyperspectral image (HSI) classification task, each pixel is categorized into a specific land-cover category or material. Convolutional neural networks (CNNs) and transformers have been widely used to extract local and non-local features in HSI classification. Recent works have utilized a multi-scale vision transformer (ViT) to enhance spectral feature capture and yield promising results. However, most existing methods still face challenges in the effective joint use of spatial-spectral information and in preserving information across layers during the propagation process. To address these issues, we propose a synergistic CNN-Transformer network with pooling attention fusion for HSI classification, which collaboratively utilizes CNNs and ViT to process spatial and spectral features separately. Specifically, we propose a Twin-Branch Feature Extraction (TBFE) module, which employs 3D and 2D convolution in parallel to comprehensively extract spectral and spatial features from HSI. A hybrid pooling attention (HPA) module is designed to aggregate spatial attention. Moreover, a cascade transformer encoder is employed for global spectral feature extraction, and a simple yet efficient cross-layer feature fusion (CFF) module is designed to reduce the loss of crucial information in the previous network layers. Extensive experiments are conducted on several representative datasets to demonstrate the superior performance of our proposed method compared to the state-of-the-art works. Code is available at https://github.com/chenpeng052/SCT-Net.git.

Zongxia Li, Xiyang Wu, Guangyao Shi et al.

Vision-Language Models (VLMs) have achieved strong results in video understanding, yet a key question remains: do they truly comprehend visual content or only learn shallow correlations between vision and language? Real visual understanding, especially of physics and common sense, is essential for AI systems that interact with the physical world. Current evaluations mostly use real-world videos similar to training data, so high benchmark scores may not reflect real reasoning ability. To address this, we propose negative-control tests using videos that depict physically impossible or logically inconsistent events. We introduce VideoHallu, a synthetic dataset of physics- and commonsense-violating scenes generated with Veo2, Sora, and Kling. It includes expert-annotated question-answer pairs across four categories of violations. Tests of leading VLMs (Qwen-2.5-VL, Video-R1, VideoChat-R1) show that, despite strong results on benchmarks such as MVBench and MMVU, they often miss these violations, exposing gaps in visual reasoning. Reinforcement learning fine-tuning on VideoHallu improves recognition of such violations without reducing standard benchmark performance. Our data is available at https://github.com/zli12321/VideoHallu.git.

Zhehui Huang, Guangyao Shi, Gaurav S. Sukhatme

Routing problems are common in mobile robotics, encompassing tasks such as inspection, surveillance, and coverage. Depending on the objective and constraints, these problems often reduce to variants of the Traveling Salesman Problem (TSP), with solutions traditionally derived by translating high-level objectives into an optimization formulation and using modern solvers to arrive at a solution. Here, we explore the potential of Large Language Models (LLMs) to replace the entire pipeline from tasks described in natural language to the generation of robot routes. We systematically investigate the performance of LLMs in robot routing by constructing a dataset with 80 unique robot routing problems across 8 variants in both single and multi-robot settings. We evaluate LLMs through three frameworks: single attempt, self-debugging, and self-debugging with self-verification and various contexts, including mathematical formulations, pseudo-code, and related research papers. Our findings reveal that both self-debugging and self-verification enhance success rates without significantly lowering the optimality gap. We observe context-sensitive behavior - providing mathematical formulations as context decreases the optimality gap but significantly decreases success rates and providing pseudo-code and related research papers as context does not consistently improve success rates or decrease the optimality gap. We identify key challenges and propose future directions to enhance LLM performance in solving robot routing problems. Our source code is available on the project website: https://sites.google.com/view/words-to-routes/.

Guangyao Shi, Yuwei Wu, Vijay Kumar et al.

Enabling robot teams to execute natural language commands requires translating high-level instructions into feasible, efficient multi-robot plans. While Large Language Models (LLMs) combined with Planning Domain Description Language (PDDL) offer promise for single-robot scenarios, existing approaches struggle with multi-robot coordination due to brittle task decomposition, poor scalability, and low coordination efficiency. We introduce PIP-LLM, a language-based coordination framework that consists of PDDL-based team-level planning and Integer Programming (IP) based robot-level planning. PIP-LLMs first decomposes the command by translating the command into a team-level PDDL problem and solves it to obtain a team-level plan, abstracting away robot assignment. Each team-level action represents a subtask to be finished by the team. Next, this plan is translated into a dependency graph representing the subtasks' dependency structure. Such a dependency graph is then used to guide the robot-level planning, in which each subtask node will be formulated as an IP-based task allocation problem, explicitly optimizing travel costs and workload while respecting robot capabilities and user-defined constraints. This separation of planning from assignment allows PIP-LLM to avoid the pitfalls of syntax-based decomposition and scale to larger teams. Experiments across diverse tasks show that PIP-LLM improves plan success rate, reduces maximum and average travel costs, and achieves better load balancing compared to state-of-the-art baselines.

Jingxi Chen, Zongxia Li, Zhichao Liu et al.

What role does the first frame play in video generation models? Traditionally, it's viewed as the spatial-temporal starting point of a video, merely a seed for subsequent animation. In this work, we reveal a fundamentally different perspective: video models implicitly treat the first frame as a conceptual memory buffer that stores visual entities for later reuse during generation. Leveraging this insight, we show that it's possible to achieve robust and generalized video content customization in diverse scenarios, using only 20-50 training examples without architectural changes or large-scale finetuning. This unveils a powerful, overlooked capability of video generation models for reference-based video customization.

Xiyang Wu, Zongxia Li, Jihui Jin et al.

Vision Language Models (VLMs) perform well on standard video tasks but struggle with physics-driven reasoning involving motion dynamics and spatial interactions. This limitation reduces their ability to interpret real or AI-generated content (AIGC) videos and to generate physically consistent content. We present an approach that addresses this gap by translating physical-world context cues into interpretable representations aligned with VLMs' perception, comprehension, and reasoning. We introduce MASS-Bench, a comprehensive benchmark consisting of 4,350 real-world and AIGC videos and 8,361 free-form video question-answering pairs focused on physics-related comprehension tasks, with detailed annotations including visual detections, sub-segment grounding, and full-sequence 3D motion tracking of entities. We further present MASS, a model-agnostic method that injects spatial-temporal signals into the VLM language space via depth-based 3D encoding and visual grounding, coupled with a motion tracker for object dynamics. To strengthen cross-modal alignment and reasoning, we apply reinforcement fine-tuning. Experiments and ablations show that our refined VLMs outperform comparable and larger baselines, as well as prior state-of-the-art models, by 8.7% and 6.0%, achieving performance comparable to close-source SoTA VLMs such as Gemini-2.5-Flash on physics reasoning and comprehension. These results validate the effectiveness of our approach.

Chak Lam Shek, Guangyao Shi, Pratap Tokekar

Multi-agent reinforcement learning (MARL) requires coordinated and stable policy updates among interacting agents. Heterogeneous-Agent Trust Region Policy Optimization (HATRPO) enforces per-agent trust region constraints using Kullback-Leibler (KL) divergence to stabilize training. However, assigning each agent the same KL threshold can lead to slow and locally optimal updates, especially in heterogeneous settings. To address this limitation, we propose two approaches for allocating the KL divergence threshold across agents: HATRPO-W, a Karush-Kuhn-Tucker-based (KKT-based) method that optimizes threshold assignment under global KL constraints, and HATRPO-G, a greedy algorithm that prioritizes agents based on improvement-to-divergence ratio. By connecting sequential policy optimization with constrained threshold scheduling, our approach enables more flexible and effective learning in heterogeneous-agent settings. Experimental results demonstrate that our methods significantly boost the performance of HATRPO, achieving faster convergence and higher final rewards across diverse MARL benchmarks. Specifically, HATRPO-W and HATRPO-G achieve comparable improvements in final performance, each exceeding 22.5%. Notably, HATRPO-W also demonstrates more stable learning dynamics, as reflected by its lower variance.

Zhehui Huang, Guangyao Shi, Yuwei Wu et al.

Multi-robot coordination has traditionally relied on a mission-specific and expert-driven pipeline, where natural language mission descriptions are manually translated by domain experts into mathematical formulation, algorithm design, and executable code. This conventional process is labor-intensive, inaccessible to non-experts, and inflexible to changes in mission requirements. Here, we propose LAN2CB (Language to Collective Behavior), a novel framework that leverages large language models (LLMs) to streamline and generalize the multi-robot coordination pipeline. LAN2CB transforms natural language (NL) mission descriptions into executable Python code for multi-robot systems through two core modules: (1) Mission Analysis, which parses mission descriptions into behavior trees, and (2) Code Generation, which leverages the behavior tree and a structured knowledge base to generate robot control code. We further introduce a dataset of natural language mission descriptions to support development and benchmarking. Experiments in both simulation and real-world environments demonstrate that LAN2CB enables robust and flexible multi-robot coordination from natural language, significantly reducing manual engineering effort and supporting broad generalization across diverse mission types. Website: https://sites.google.com/view/lan-cb

Guangyao Shi, Ishat E Rabban, Lifeng Zhou et al.

Submodular maximization has been widely used in many multi-robot task planning problems including information gathering, exploration, and target tracking. However, the interplay between submodular maximization and communication is rarely explored in the multi-robot setting. In many cases, maximizing the submodular objective may drive the robots in a way so as to disconnect the communication network. Driven by such observations, in this paper, we consider the problem of maximizing submodular function with connectivity constraints. Specifically, we propose a problem called Communication-aware Submodular Maximization (CSM), in which communication maintenance and submodular maximization are jointly considered in the decision-making process. One heuristic algorithm that consists of two stages, i.e. \textit{topology generation} and \textit{deviation minimization} is proposed. We validate the formulation and algorithm through numerical simulation. We find that our algorithm on average suffers only slightly performance decrease compared to the pure greedy strategy.

Md Ishat-E-Rabban, Guangyao Shi, Pratap Tokekar

Maintaining a robust communication network plays an important role in the success of a multi-robot team jointly performing an optimization task. A key characteristic of a robust multi-robot system is the ability to repair the communication topology itself in the case of robot failure. In this paper, we focus on the Fast Biconnectivity Restoration (FBR) problem, which aims to repair a connected network to make it biconnected as fast as possible, where a biconnected network is a communication topology that cannot be disconnected by removing one node. We develop a Quadratically Constrained Program (QCP) formulation of the FBR problem, which provides a way to optimally solve the problem. We also propose an approximation algorithm for the FBR problem based on graph theory. By conducting empirical studies, we demonstrate that our proposed approximation algorithm performs close to the optimal while significantly outperforming the existing solutions.

Guangyao Shi, Lifeng Zhou, Pratap Tokekar

The multiple-path orienteering problem asks for paths for a team of robots that maximize the total reward collected while satisfying budget constraints on the path length. This problem models many multi-robot routing tasks such as exploring unknown environments and information gathering for environmental monitoring. In this paper, we focus on how to make the robot team robust to failures when operating in adversarial environments. We introduce the Robust Multiple-path Orienteering Problem (RMOP) where we seek worst-case guarantees against an adversary that is capable of attacking at most $α$ robots. We consider two versions of this problem: RMOP offline and RMOP online. In the offline version, there is no communication or replanning when robots execute their plans and our main contribution is a general approximation scheme with a bounded approximation guarantee that depends on $α$ and the approximation factor for single robot orienteering. In particular, we show that the algorithm yields a (i) constant-factor approximation when the cost function is modular; (ii) $\log$ factor approximation when the cost function is submodular; and (iii) constant-factor approximation when the cost function is submodular but the robots are allowed to exceed their path budgets by a bounded amount. In the online version, RMOP is modeled as a two-player sequential game and solved adaptively in a receding horizon fashion based on Monte Carlo Tree Search (MCTS). In addition to theoretical analysis, we perform simulation studies for ocean monitoring and tunnel information-gathering applications to demonstrate the efficacy of our approach.

Hong Huang, Guangyao Shi, Haibo He et al.

The graph embedding (GE) methods have been widely applied for dimensionality reduction of hyperspectral imagery (HSI). However, a major challenge of GE is how to choose proper neighbors for graph construction and explore the spatial information of HSI data. In this paper, we proposed an unsupervised dimensionality reduction algorithm termed spatial-spectral manifold reconstruction preserving embedding (SSMRPE) for HSI classification. At first, a weighted mean filter (WMF) is employed to preprocess the image, which aims to reduce the influence of background noise. According to the spatial consistency property of HSI, the SSMRPE method utilizes a new spatial-spectral combined distance (SSCD) to fuse the spatial structure and spectral information for selecting effective spatial-spectral neighbors of HSI pixels. Then, it explores the spatial relationship between each point and its neighbors to adjusts the reconstruction weights for improving the efficiency of manifold reconstruction. As a result, the proposed method can extract the discriminant features and subsequently improve the classification performance of HSI. The experimental results on PaviaU and Salinas hyperspectral datasets indicate that SSMRPE can achieve better classification accuracies in comparison with some state-of-the-art methods.