arXiv subjects

Haoyu Fu

Publications and source records attributed to Haoyu Fu.

8 recordsLinked to original sources

arXiv2026

Event-Based Upper-Body Humanoid Teleoperation Under Challenging Illumination

We present a real-time upper-body human-to-humanoid motion imitation framework driven by neuromorphic event-based vision. This work addresses practical perceptual bottlenecks of conventional frame-based RGB sensors, specifically their difficulty in high dynamic range (HDR) scenes and rapid motions due to fixed integration times. By leveraging the Prophesee EVK4 event camera, which operates asynchronously with high temporal resolution and a dynamic range exceeding 120 dB, our system supports stable tracking in conditions where standard vision pipelines degrade, such as severe backlighting and very low light environments below 5 lux. The architecture integrates a low-latency Perception Module, utilizing optimized event accumulation and gravity-aligned inertial fusion, with a causal Motion Module (TWIST) that performs online kinematic retargeting. We validate the system on an embedded NVIDIA Booster T1 platform and an 18-DoF humanoid upper-body setup, demonstrating an end-to-end photon-to-action latency of 23-34 ms and advantages over RGB baselines under our experimental setup. The results indicate a practical trade-off: events can be preferable for fast or poorly lit upper-body teleoperation, whereas well-lit static scenes may favor RGB or hybrid sensing.

cs.RO↗

arXiv2026

JADE-GS: Joint Allocation of Deblurring Evidence for Event-Assisted 3D Gaussian Splatting

Neural radiance fields and 3D Gaussian Splatting assume that each training image is a sharp and geometrically consistent observation of the scene. Motion blur violates this assumption because a single exposure integrates a continuous range of camera poses. Exposure integration also removes the temporal information needed to recover the corresponding sharp observation. Event cameras preserve this information at microsecond resolution and therefore provide a natural complement to conventional images. Existing event-assisted reconstruction methods predominantly obtain image supervision through analytical inversion of the Event Double Integral. Learned restoration from frames and events offers a second prior. Although weaker when used alone, it fails in different regions and provides complementary evidence. We present JADE-GS, which formulates the combination of these priors as spatial evidence allocation. A lightweight Spatial Prior Router predicts a pixelwise allocation using only the blurry frame and event stream, then fuses the two fixed restorations into an additional supervision target. The router is trained without a sharp reference using consistency with the scene under reconstruction and the measured exposure, and is removed after optimization. Experiments show that JADE-GS achieves leading perceptual quality on both benchmarks, attains the best fidelity on the real benchmark, and remains competitive on the synthetic one. It requires substantially lower training overhead than diffusion-based alternatives and preserves native 3DGS rendering with no generative decoding at inference.

cs.CV↗

arXiv2025

MindDrive: A Vision-Language-Action Model for Autonomous Driving via Online Reinforcement Learning

Current Vision-Language-Action (VLA) paradigms in autonomous driving primarily rely on Imitation Learning (IL), which introduces inherent challenges such as distribution shift and causal confusion. Online Reinforcement Learning offers a promising pathway to address these issues through trial-and-error learning. However, applying online reinforcement learning to VLA models in autonomous driving is hindered by inefficient exploration in continuous action spaces. To overcome this limitation, we propose MindDrive, a VLA framework comprising a large language model (LLM) with two distinct sets of LoRA parameters. The one LLM serves as a Decision Expert for scenario reasoning and driving decision-making, while the other acts as an Action Expert that dynamically maps linguistic decisions into feasible trajectories. By feeding trajectory-level rewards back into the reasoning space, MindDrive enables trial-and-error learning over a finite set of discrete linguistic driving decisions, instead of operating directly in a continuous action space. This approach effectively balances optimal decision-making in complex scenarios, human-like driving behavior, and efficient exploration in online reinforcement learning. Using the lightweight Qwen-0.5B LLM, MindDrive achieves Driving Score (DS) of 78.04 and Success Rate (SR) of 55.09% on the challenging Bench2Drive benchmark. To the best of our knowledge, this is the first work to demonstrate the effectiveness of online reinforcement learning for the VLA model in autonomous driving.

cs.CV↗

arXiv2025

Extending Large Vision-Language Model for Diverse Interactive Tasks in Autonomous Driving

The Large Visual-Language Models (LVLMs) have significantly advanced image understanding. Their comprehension and reasoning capabilities enable promising applications in autonomous driving scenarios. However, existing research typically focuses on front-view perspectives and partial objects within scenes, struggling to achieve comprehensive scene understanding. Meanwhile, existing LVLMs suffer from the lack of mapping relationship between 2D and 3D and insufficient integration of 3D object localization and instruction understanding. To tackle these limitations, we first introduce NuInteract, a large-scale dataset with over 1.5M multi-view image language pairs spanning dense scene captions and diverse interactive tasks. Furthermore, we propose DriveMonkey, a simple yet effective framework that seamlessly integrates LVLMs with a spatial processor using a series of learnable queries. The spatial processor, designed as a plug-and-play component, can be initialized with pre-trained 3D detectors to improve 3D perception. Our experiments show that DriveMonkey outperforms general LVLMs, especially achieving a 9.86% notable improvement on the 3D visual grounding task. The dataset and code will be released at https://github.com/zc-zhao/DriveMonkey.

cs.CV↗

arXiv2025

ORION: A Holistic End-to-End Autonomous Driving Framework by Vision-Language Instructed Action Generation

End-to-end (E2E) autonomous driving methods still struggle to make correct decisions in interactive closed-loop evaluation due to limited causal reasoning capability. Current methods attempt to leverage the powerful understanding and reasoning abilities of Vision-Language Models (VLMs) to resolve this dilemma. However, the problem is still open that few VLMs for E2E methods perform well in the closed-loop evaluation due to the gap between the semantic reasoning space and the purely numerical trajectory output in the action space. To tackle this issue, we propose ORION, a holistic E2E autonomous driving framework by vision-language instructed action generation. ORION uniquely combines a QT-Former to aggregate long-term history context, a Large Language Model (LLM) for driving scenario reasoning, and a generative planner for precision trajectory prediction. ORION further aligns the reasoning space and the action space to implement a unified E2E optimization for both visual question-answering (VQA) and planning tasks. Our method achieves an impressive closed-loop performance of 77.74 Driving Score (DS) and 54.62% Success Rate (SR) on the challenge Bench2Drive datasets, which outperforms state-of-the-art (SOTA) methods by a large margin of 14.28 DS and 19.61% SR.

cs.CV↗

arXiv2018

Guaranteed Recovery of One-Hidden-Layer Neural Networks via Cross Entropy

We study model recovery for data classification, where the training labels are generated from a one-hidden-layer neural network with sigmoid activations, also known as a single-layer feedforward network, and the goal is to recover the weights of the neural network. We consider two network models, the fully-connected network (FCN) and the non-overlapping convolutional neural network (CNN). We prove that with Gaussian inputs, the empirical risk based on cross entropy exhibits strong convexity and smoothness {\em uniformly} in a local neighborhood of the ground truth, as soon as the sample complexity is sufficiently large. This implies that if initialized in this neighborhood, gradient descent converges linearly to a critical point that is provably close to the ground truth. Furthermore, we show such an initialization can be obtained via the tensor method. This establishes the global convergence guarantee for empirical risk minimization using cross entropy via gradient descent for learning one-hidden-layer neural networks, at the near-optimal sample and computational complexity with respect to the network input dimension without unrealistic assumptions such as requiring a fresh set of samples at each iteration.

stat.ML↗

arXiv2017

Quantized Spectral Compressed Sensing: Cramer-Rao Bounds and Recovery Algorithms

Efficient estimation of wideband spectrum is of great importance for applications such as cognitive radio. Recently, sub-Nyquist sampling schemes based on compressed sensing have been proposed to greatly reduce the sampling rate. However, the important issue of quantization has not been fully addressed, particularly for high-resolution spectrum and parameter estimation. In this paper, we aim to recover spectrally-sparse signals and the corresponding parameters, such as frequency and amplitudes, from heavy quantizations of their noisy complex-valued random linear measurements, e.g. only the quadrant information. We first characterize the Cramer-Rao bound under Gaussian noise, which highlights the trade-off between sample complexity and bit depth under different signal-to-noise ratios for a fixed budget of bits. Next, we propose a new algorithm based on atomic norm soft thresholding for signal recovery, which is equivalent to proximal mapping of properly designed surrogate signals with respect to the atomic norm that motivates spectral sparsity. The proposed algorithm can be applied to both the single measurement vector case, as well as the multiple measurement vector case. It is shown that under the Gaussian measurement model, the spectral signals can be reconstructed accurately with high probability, as soon as the number of quantized measurements exceeds the order of K log n, where K is the level of spectral sparsity and $n$ is the signal dimension. Finally, numerical simulations are provided to validate the proposed approaches.

eess.SP↗

arXiv2014

Subspace Learning From Bits

Networked sensing, where the goal is to perform complex inference using a large number of inexpensive and decentralized sensors, has become an increasingly attractive research topic due to its applications in wireless sensor networks and internet-of-things. To reduce the communication, sensing and storage complexity, this paper proposes a simple sensing and estimation framework to faithfully recover the principal subspace of high-dimensional data streams using a collection of binary measurements from distributed sensors, without transmitting the whole data. The binary measurements are designed to indicate comparison outcomes of aggregated energy projections of the data samples over pairs of randomly selected directions. When the covariance matrix is a low-rank matrix, we propose a spectral estimator that recovers the principal subspace of the covariance matrix as the subspace spanned by the top eigenvectors of a properly designed surrogate matrix, which is provably accurate as soon as the number of binary measurements is sufficiently large. An adaptive rank selection strategy based on soft thresholding is also presented. Furthermore, we propose a tailored spectral estimator when the covariance matrix is additionally Toeplitz, and show reliable estimation can be obtained from a substantially smaller number of binary measurements. Our results hold even when a constant fraction of the binary measurements is randomly flipped. Finally, we develop a low-complexity online algorithm to track the principal subspace when new measurements arrive sequentially. Numerical examples are provided to validate the proposed approach.

stat.ML↗