arXiv subjects

Mengzhen Liu

Publications and source records attributed to Mengzhen Liu.

At least 19 recordsLinked to original sources

arXiv2026

SaPaVe: Towards Active Perception and Manipulation in Vision-Language-Action Models for Robotics

Active perception and manipulation are crucial for robots to interact with complex scenes. Existing methods struggle to unify semantic-driven active perception with robust, viewpoint-invariant execution. We propose SaPaVe, an end-to-end framework that jointly learns these capabilities in a data-efficient manner. Our approach decouples camera and manipulation actions rather than placing them in a shared action space, and follows a bottom-up training strategy: we first train semantic camera control on a large-scale dataset, then jointly optimize both action types using hybrid data. To support this framework, we introduce ActiveViewPose-200K, a dataset of 200k image-language-camera movement pairs for semantic camera movement learning, and a 3D geometry-aware module that improves execution robustness under dynamic viewpoints. We also present ActiveManip-Bench, the first benchmark for evaluating active manipulation beyond fixed-view settings. Extensive experiments in both simulation and real-world environments show that SaPaVe outperforms recent vision-language-action models such as GR00T N1 and $\pi_0$, achieving up to 31.25\% higher success rates in real-world tasks. These results show that tightly coupled perception and execution, when trained with decoupled yet coordinated strategies, enable efficient and generalizable active manipulation. Project page: https://lmzpai.github.io/SaPaVe

cs.RO↗

arXiv2026

4-rank distribution of Picard groups of hyperelliptic curves via $C$-symmetric matrices

We determine the large-genus limiting distribution of the 4-rank of the Picard group of hyperelliptic curves over a fixed finite field $\mathbb F_q$ of odd characteristic. This is a function field analogue of a result of Fouvry and Kl\"uners. Our computation agrees with (the Picard group analogue of) the Cohen--Lenstra--Gerth heuristics in the case $q \equiv 3\pmod{4}$, i.e., in the absence of roots of unity in the base field. When roots of unity are present, the result is of the same form as conjectured distribution for class groups of quadratic extensions of number fields containing roots of unity. The limiting distribution does not change when imposing finitely many conditions on the ramification behavior of the curves. In the process, we determine the rank distribution of a certain class of random matrix ensembles over finite fields determined by symmetry conditions.

math.NT↗

arXiv2026

RoboBrain 2.5: Depth in Sight, Time in Mind

We introduce RoboBrain 2.5, a next-generation embodied AI foundation model that advances general perception, spatial reasoning, and temporal modeling through extensive training on high-quality spatiotemporal supervision. Building upon its predecessor, RoboBrain 2.5 introduces two major capability upgrades. Specifically, it unlocks Precise 3D Spatial Reasoning by shifting from 2D pixel-relative grounding to depth-aware coordinate prediction and absolute metric constraint comprehension, generating complete 3D manipulation traces as ordered keypoint sequences under physical constraints. Complementing this spatial precision, the model establishes Dense Temporal Value Estimation that provides dense, step-aware progress prediction and execution state understanding across varying viewpoints, producing stable feedback signals for downstream learning. Together, these upgrades extend the framework toward more physically grounded and execution-aware embodied intelligence for complex, fine-grained manipulation. The code and checkpoints are available at project website: https://superrobobrain.github.io

cs.RO↗

arXiv2025

Towards Spatial Trace with Reasoning in Vision-Language Models for Robotics

Spatial tracing, as a fundamental embodied interaction ability for robots, is inherently challenging as it requires multi-step metric-grounded reasoning compounded with complex spatial referring and real-world metric measurement. However, existing methods struggle with this compositional task. To this end, we propose RoboTracer, a 3D-aware VLM that first achieves both 3D spatial referring and measuring via a universal spatial encoder and a regression-supervised decoder to enhance scale awareness during supervised fine-tuning (SFT). Moreover, RoboTracer advances multi-step metric-grounded reasoning via reinforcement fine-tuning (RFT) with metric-sensitive process rewards, supervising key intermediate perceptual cues to accurately generate spatial traces. To support SFT and RFT training, we introduce TraceSpatial, a large-scale dataset of 30M QA pairs, spanning outdoor/indoor/tabletop scenes and supporting complex reasoning processes (up to 9 steps). We further present TraceSpatial-Bench, a challenging benchmark filling the gap to evaluate spatial tracing. Experimental results show that RoboTracer surpasses baselines in spatial understanding, measuring, and referring, with an average success rate of 79.1%, and also achieves SOTA performance on TraceSpatial-Bench by a large margin, exceeding Gemini-2.5-Pro by 36% accuracy. Notably, RoboTracer can be integrated with various control policies to execute long-horizon, dynamic tasks across diverse robots (UR5, G1 humanoid) in cluttered real-world scenes. Please see the project page at https://zhoues.github.io/RoboTracer.

cs.RO↗

arXiv2025

Reconfigurable Antenna Arrays: Bridging Electromagnetics and Signal Processing

Reconfigurable antennas (RAs), capable of dynamically adapting their radiation patterns, polarization states, and operating frequencies, have emerged as a promising technology to meet the stringent performance requirements of sixth-generation (6G) wireless networks. This article systematically introduces essential hardware implementations of RAs and investigates advanced array architectures, such as fully-digital and tri-hybrid designs, emphasizing their capability to synergistically integrate electromagnetic (EM) reconfigurability with analog and digital signal processing. By facilitating coordinated beamforming across the EM and signal processing domains, RA arrays offer unprecedented flexibility and adaptability compared to conventional static antenna systems. Representative applications empowered by RA arrays, including integrated sensing and communication (ISAC), physical layer security (PLS), and near-field communications, are highlighted. A case study illustrates the effectiveness of RA arrays in optimizing beam steering, improving link robustness, and alleviating system power consumption. Finally, several open challenges and future research directions are outlined, emphasizing the need for advancements in theoretical modeling, hardware reliability, channel estimation techniques, intelligent optimization methods, and innovative network architectures, to fully realize the transformative impact of RAs in future 6G wireless networks.

eess.SP↗

arXiv2025

MotionTrans: Human VR Data Enable Motion-Level Learning for Robotic Manipulation Policies

Scaling real robot data is a key bottleneck in imitation learning, leading to the use of auxiliary data for policy training. While other aspects of robotic manipulation such as image or language understanding may be learned from internet-based datasets, acquiring motion knowledge remains challenging. Human data, with its rich diversity of manipulation behaviors, offers a valuable resource for this purpose. While previous works show that using human data can bring benefits, such as improving robustness and training efficiency, it remains unclear whether it can realize its greatest advantage: enabling robot policies to directly learn new motions for task completion. In this paper, we systematically explore this potential through multi-task human-robot cotraining. We introduce MotionTrans, a framework that includes a data collection system, a human data transformation pipeline, and a weighted cotraining strategy. By cotraining 30 human-robot tasks simultaneously, we direcly transfer motions of 13 tasks from human data to deployable end-to-end robot policies. Notably, 9 tasks achieve non-trivial success rates in zero-shot manner. MotionTrans also significantly enhances pretraining-finetuning performance (+40% success rate). Through ablation study, we also identify key factors for successful motion learning: cotraining with robot data and broad task-related motion coverage. These findings unlock the potential of motion-level learning from human data, offering insights into its effective use for training robotic manipulation policies. All data, code, and model weights are open-sourced https://motiontrans.github.io/.

cs.RO↗

arXiv2025

Clutter Suppression in ISAC Systems with Compound Reconfigurable Antenna Arrays

Integrated sensing and communication (ISAC) systems often suffer severe performance degradation due to strong clutter echoes, and spatial-only beamforming is often inadequate for realistic array sizes. This paper addresses clutter suppression in ISAC by leveraging compound reconfigurable antenna (CRA) arrays, which simultaneously enable dynamic adjustment of both radiation patterns and polarization states, thus substantially expanding the degrees of freedom available in the electromagnetic (EM) domain. We develop a unified compound channel model that integrates virtual angular-domain responses, spatial propagation, and polarization rotation/depolarization. Leveraging statistical information about target and clutter covariances, we formulate a joint EM-domain and baseband-domain optimization aimed at maximizing the radar signal-to-clutter-plus-noise ratio (SCNR). The formulation also enforces multiuser downlink signal-to-interference-plus-noise ratio constraints, a total transmit-power budget, and finite-codebook EM-mode selection. The resulting nonconvex mixed-integer problem is tackled by an alternating algorithm that combines fractional programming and majorization-minimization with second-order cone programming-based updates and a penalty relaxation for mode selection. Extensive simulations in QuaDRiGa-based channel environments validate the effectiveness of the proposed CRA array design, demonstrating up to 11 dB SCNR improvements over conventional beamforming methods relying solely on baseband-domain optimization and confirming the substantial benefits of fully exploiting EM-domain reconfigurability for clutter-rich ISAC scenarios.

eess.SP↗

arXiv2025

Beyond Attention or Similarity: Maximizing Conditional Diversity for Token Pruning in MLLMs

In multimodal large language models (MLLMs), the length of input visual tokens is often significantly greater than that of their textual counterparts, leading to a high inference cost. Many works aim to address this issue by removing redundant visual tokens. However, current approaches either rely on attention-based pruning, which retains numerous duplicate tokens, or use similarity-based pruning, overlooking the instruction relevance, consequently causing suboptimal performance. In this paper, we go beyond attention or similarity by proposing a novel visual token pruning method named CDPruner, which maximizes the conditional diversity of retained tokens. We first define the conditional similarity between visual tokens conditioned on the instruction, and then reformulate the token pruning problem with determinantal point process (DPP) to maximize the conditional diversity of the selected subset. The proposed CDPruner is training-free and model-agnostic, allowing easy application to various MLLMs. Extensive experiments across diverse MLLMs show that CDPruner establishes new state-of-the-art on various vision-language benchmarks. By maximizing conditional diversity through DPP, the selected subset better represents the input images while closely adhering to user instructions, thereby preserving strong performance even with high reduction ratios. When applied to LLaVA, CDPruner reduces FLOPs by 95\% and CUDA latency by 78\%, while maintaining 94\% of the original accuracy. Our code is available at https://github.com/Theia-4869/CDPruner.

cs.CV↗

arXiv2025

HybridVLA: Collaborative Diffusion and Autoregression in a Unified Vision-Language-Action Model

A fundamental objective of manipulation policy design is to endow robots to comprehend human instructions, reason about scene cues, and execute generalized actions in dynamic environments. Recent autoregressive vision-language-action (VLA) methods inherit common-sense reasoning capabilities from vision-language models (VLMs) for next action-token prediction. However, these methods quantize actions into discrete bins, which disrupts the continuity required for precise control. In contrast, existing diffusion-based VLA methods incorporate an additional diffusion head to predict continuous actions solely conditioned on feature representations extracted by the VLM, without fully leveraging the VLM's pretrained reasoning capabilities through token-level generation. To address these limitations, we introduce HybridVLA, a unified framework that absorbs the continuous nature of diffusion-based actions and the contextual reasoning of autoregression within a single large language model. To mitigate interference between the two generation paradigms, we propose a collaborative training recipe that seamlessly incorporates diffusion denoising into the next-token prediction process. With this recipe, we find these two action prediction methods not only reinforce each other but also exhibit varying strength across different tasks. Therefore, we design a collaborative action ensemble mechanism that adaptively fuses both predictions, leading to more robust control. HybridVLA outperforms previous state-of-the-art VLA methods by 14\% and 19\% in mean success rate on simulation and real-world tasks, respectively, while demonstrating stable manipulation in unseen configurations.

cs.CV↗

arXiv2025

Distributed Distortion-Aware Beamforming Designs for Cell-Free mMIMO Systems

Cell-free massive multi-input multi-output (CF-mMIMO) systems have emerged as a promising paradigm for next-generation wireless communications, offering enhanced spectral efficiency and coverage through distributed antenna arrays. However, the non-linearity of power amplifiers (PAs) in these arrays introduce spatial distortion, which may significantly degrade system performance. This paper presents the first investigation of distortion-aware beamforming in a distributed framework tailored for CF-mMIMO systems, enabling pre-compensation for beam dispersion caused by nonlinear PA distortion. Using a third-order memoryless polynomial distortion model, the impact of the nonlinear PA on the performance of CF-mMIMO systems is firstly analyzed by evaluating the signal-to-interference-noise-and-distortion ratio (SINDR) at user equipment (UE). Then, we develop two distributed distortion-aware beamforming designs based on ring topology and star topology, respectively. In particular, the ring-topology-based fully-distributed approach reduces interconnection costs and computational complexity, while the star-topology-based partially-distributed scheme leverages the superior computation capability of the central processor to achieve improved sum-rate performance. Extensive simulations demonstrate the effectiveness of the proposed distortion-aware beamforming designs in mitigating the effect of nonlinear PA distortion, while also reducing computational complexity and backhaul information exchange in CF-mMIMO systems.

eess.SP↗

arXiv2025

Tri-timescale Beamforming Design for Tri-hybrid Architectures with Reconfigurable Antennas

Reconfigurable antennas possess the capability to dynamically adjust their fundamental operating characteristics, thereby enhancing system adaptability and performance. To fully exploit this flexibility in modern wireless communication systems, this paper considers a novel tri-hybrid beamforming architecture, which seamlessly integrates pattern-reconfigurable antennas with both analog and digital beamforming. The proposed tri-hybrid architecture operates across three layers: (\textit{i}) a radiation beamformer in the electromagnetic (EM) domain for dynamic pattern alignment, (\textit{ii}) an analog beamformer in the radio-frequency (RF) domain for array gain enhancement, and (\textit{iii}) a digital beamformer in the baseband (BB) domain for multi-user interference mitigation. To establish a solid theoretical foundation, we first develop a comprehensive mathematical model for the tri-hybrid beamforming system and formulate the signal model for a multi-user multi-input single-output (MU-MISO) scenario. The optimization objective is to maximize the sum-rate while satisfying practical constraints. Given the challenges posed by high pilot overhead and computational complexity, we introduce an innovative tri-timescale beamforming framework, wherein the radiation beamformer is optimized over a long-timescale, the analog beamformer over a medium-timescale, and the digital beamformer over a short-timescale. This hierarchical strategy effectively balances performance and implementation feasibility. Simulation results validate the performance gains of the proposed tri-hybrid architecture and demonstrate that the tri-timescale design significantly reduces pilot overhead and computational complexity, highlighting its potential for future wireless communication systems.

eess.SP↗

arXiv2025

CordViP: Correspondence-based Visuomotor Policy for Dexterous Manipulation in Real-World

Achieving human-level dexterity in robots is a key objective in the field of robotic manipulation. Recent advancements in 3D-based imitation learning have shown promising results, providing an effective pathway to achieve this goal. However, obtaining high-quality 3D representations presents two key problems: (1) the quality of point clouds captured by a single-view camera is significantly affected by factors such as camera resolution, positioning, and occlusions caused by the dexterous hand; (2) the global point clouds lack crucial contact information and spatial correspondences, which are necessary for fine-grained dexterous manipulation tasks. To eliminate these limitations, we propose CordViP, a novel framework that constructs and learns correspondences by leveraging the robust 6D pose estimation of objects and robot proprioception. Specifically, we first introduce the interaction-aware point clouds, which establish correspondences between the object and the hand. These point clouds are then used for our pre-training policy, where we also incorporate object-centric contact maps and hand-arm coordination information, effectively capturing both spatial and temporal dynamics. Our method demonstrates exceptional dexterous manipulation capabilities, achieving state-of-the-art performance in six real-world tasks, surpassing other baselines by a large margin. Experimental results also highlight the superior generalization and robustness of CordViP to different objects, viewpoints, and scenarios. Code and videos are available on https://aureleopku.github.io/CordViP.

cs.RO↗

arXiv2024

RoboMIND: Benchmark on Multi-embodiment Intelligence Normative Data for Robot Manipulation

In this paper, we introduce RoboMIND (Multi-embodiment Intelligence Normative Data for Robot Manipulation), a dataset containing 107k demonstration trajectories across 479 diverse tasks involving 96 object classes. RoboMIND is collected through human teleoperation and encompasses comprehensive robotic-related information, including multi-view observations, proprioceptive robot state information, and linguistic task descriptions. To ensure data consistency and reliability for imitation learning, RoboMIND is built on a unified data collection platform and a standardized protocol, covering four distinct robotic embodiments: the Franka Emika Panda, the UR5e, the AgileX dual-arm robot, and a humanoid robot with dual dexterous hands. Our dataset also includes 5k real-world failure demonstrations, each accompanied by detailed causes, enabling failure reflection and correction during policy learning. Additionally, we created a digital twin environment in the Isaac Sim simulator, replicating the real-world tasks and assets, which facilitates the low-cost collection of additional training data and enables efficient evaluation. To demonstrate the quality and diversity of our dataset, we conducted extensive experiments using various imitation learning methods for single-task settings and state-of-the-art Vision-Language-Action (VLA) models for multi-task scenarios. By leveraging RoboMIND, the VLA models achieved high manipulation success rates and demonstrated strong generalization capabilities. To the best of our knowledge, RoboMIND is the largest multi-embodiment teleoperation dataset collected on a unified platform, providing large-scale and high-quality robotic training data. Our project is at https://x-humanoid-robomind.github.io/.

cs.RO↗

arXiv2024

Dynamic Hybrid Beamforming Designs for ELAA Near-Field Communications

Extremely large-scale antenna array (ELAA) is a key candidate technology for the sixth generation (6G) mobile networks. Nevertheless, using substantial numbers of antennas to transmit high-frequency signals in ELAA systems significantly exacerbates the near-field effect. Unfortunately, traditional hybrid beamforming schemes are highly vulnerable to ELAA near-field communications. To effectively mitigate severe near-field effect, we propose a novel dynamic hybrid beamforming architecture for ELAA systems, in which each antenna is either adaptively connected to one radio frequency (RF) chain for signal transmission or deactivated for power saving. For the case that instantaneous channel state information (CSI) is available during each channel coherence time, a real-time dynamic hybrid beamforming design is developed to maximize the achievable sum rate under the constraints of the constant modulus of phase-shifters (PSs), non-overlapping dynamic connection network and total transmit power. When instantaneous CSI cannot be easily obtained in real-time, we propose a two-timescale dynamic hybrid beamforming design, which optimizes analog beamformer in long-timescale and digital beamformer in short-timescale, with the goal of maximizing ergodic sum-rate under the same constraints. Simulation results demonstrate the advantages of the proposed dynamic hybrid beamforming architecture and the effectiveness of the developed algorithms for ELAA near-field communications.

eess.SP↗

arXiv2024

RoboMamba: Efficient Vision-Language-Action Model for Robotic Reasoning and Manipulation

A fundamental objective in robot manipulation is to enable models to comprehend visual scenes and execute actions. Although existing Vision-Language-Action (VLA) models for robots can handle a range of basic tasks, they still face challenges in two areas: (1) insufficient reasoning ability to tackle complex tasks, and (2) high computational costs for VLA model fine-tuning and inference. The recently proposed state space model (SSM) known as Mamba demonstrates promising capabilities in non-trivial sequence modeling with linear inference complexity. Inspired by this, we introduce RoboMamba, an end-to-end robotic VLA model that leverages Mamba to deliver both robotic reasoning and action capabilities, while maintaining efficient fine-tuning and inference. Specifically, we first integrate the vision encoder with Mamba, aligning visual tokens with language embedding through co-training, empowering our model with visual common sense and robotic-related reasoning. To further equip RoboMamba with SE(3) pose prediction abilities, we explore an efficient fine-tuning strategy with a simple policy head. We find that once RoboMamba possesses sufficient reasoning capability, it can acquire manipulation skills with minimal fine-tuning parameters (0.1\% of the model) and time. In experiments, RoboMamba demonstrates outstanding reasoning capabilities on general and robotic evaluation benchmarks. Meanwhile, our model showcases impressive pose prediction results in both simulation and real-world experiments, achieving inference speeds 3 times faster than existing VLA models. Our project web page: https://sites.google.com/view/robomamba-web

cs.CV↗

arXiv2023

Minimisation of 2-coverings of genus 2 Jacobians

An important problem in computational arithmetic geometry is to find changes of coordinates to simplify a system of polynomial equations with rational coefficients. This is tackled by a combination of two techniques, called minimisation and reduction. We give an algorithm for minimising certain pairs of quadratic forms, subject to the constraint that the first quadratic form is fixed. This has applications to 2-descent on the Jacobian of a genus 2 curve.

math.NT↗

arXiv2022

Tiling of rectangles with squares via Diophantine approximation

This article shines new light on the classical problem of tiling rectangles with squares efficiently with a novel method. With a twist on the traditional approach of resistor networks, we provide new and improved results on the matter using the theory of Diophantine Approximation, hence overcoming long-established difficulties, such as generalizations to higher-dimensional analogues. The universality of the method is demonstrated through its applications to different tiling problems. These include tiling rectangles with other rectangles, with their respective higher-dimensional counterparts, as well as tiling equilateral triangles, parallelograms, and trapezoids with equilateral triangles.

math.CO↗

arXiv2022

Laplacian pretty good fractional revival

We develop the theory of pretty good fractional revival in quantum walks on graphs using their Laplacian matrices as the Hamiltonian. We classify the paths and the double stars that have Laplacian pretty good fractional revival.

math.CO↗