Senna: Bridging Large Vision-Language Models and End-to-End Autonomous Driving

End-to-end autonomous driving demonstrates strong planning capabilities with large-scale data but still struggles in complex, rare scenarios due to limited commonsense. In contrast, Large Vision-Language Models (LVLMs) excel in scene understanding and reasoning. The path forward lies in merging the strengths of both approaches. Previous methods using LVLMs to predict trajectories or control signals yield suboptimal results, as LVLMs are not well-suited for precise numerical predictions. This paper presents Senna, an autonomous driving system combining an LVLM (Senna-VLM) with an end-to-end model (Senna-E2E). Senna decouples high-level planning from low-level trajectory prediction. Senna-VLM generates planning decisions in natural language, while Senna-E2E predicts precise trajectories. Senna-VLM utilizes a multi-image encoding approach and multi-view prompts for efficient scene understanding. Besides, we introduce planning-oriented QAs alongside a three-stage training strategy, which enhances Senna-VLM's planning performance while preserving commonsense. Extensive experiments on two datasets show that Senna achieves state-of-the-art planning performance. Notably, with pre-training on a large-scale dataset DriveX and fine-tuning on nuScenes, Senna significantly reduces average planning error by 27.12% and collision rate by 33.33% over model without pre-training. We believe Senna's cross-scenario generalization and transferability are essential for achieving fully autonomous driving. Code and models will be released at https://github.com/hustvl/Senna.

M3Bench: Benchmarking Whole-body Motion Generation for Mobile Manipulation in 3D Scenes

We propose M^3Bench, a new benchmark of whole-body motion generation for mobile manipulation tasks. Given a 3D scene context, M^3Bench requires an embodied agent to understand its configuration, environmental constraints and task objectives, then generate coordinated whole-body motion trajectories for object rearrangement tasks. M^3Bench features 30k object rearrangement tasks across 119 diverse scenes, providing expert demonstrations generated by our newly developed M^3BenchMaker. This automatic data generation tool produces coordinated whole-body motion trajectories from high-level task instructions, requiring only basic scene and robot information. Our benchmark incorporates various task splits to assess generalization across different dimensions and leverages realistic physics simulation for trajectory evaluation. Through extensive experimental analyses, we reveal that state-of-the-art models still struggle with coordinated base-arm motion while adhering to environment-context and task-specific constraints, highlighting the need to develop new models that address this gap. Through M^3Bench, we aim to facilitate future robotics research towards more adaptive and capable mobile manipulation in diverse, real-world environments.

M2Diffuser: Diffusion-based Trajectory Optimization for Mobile Manipulation in 3D Scenes

Recent advances in diffusion models have opened new avenues for research into embodied AI agents and robotics. Despite significant achievements in complex robotic locomotion and skills, mobile manipulation-a capability that requires the coordination of navigation and manipulation-remains a challenge for generative AI techniques. This is primarily due to the high-dimensional action space, extended motion trajectories, and interactions with the surrounding environment. In this paper, we introduce M2Diffuser, a diffusion-based, scene-conditioned generative model that directly generates coordinated and efficient whole-body motion trajectories for mobile manipulation based on robot-centric 3D scans. M2Diffuser first learns trajectory-level distributions from mobile manipulation trajectories provided by an expert planner. Crucially, it incorporates an optimization module that can flexibly accommodate physical constraints and task objectives, modeled as cost and energy functions, during the inference process. This enables the reduction of physical violations and execution errors at each denoising step in a fully differentiable manner. Through benchmarking on three types of mobile manipulation tasks across over 20 scenes, we demonstrate that M2Diffuser outperforms state-of-the-art neural planners and successfully transfers the generated trajectories to a real-world robot. Our evaluations underscore the potential of generative AI to enhance the generalization of traditional planning and learning-based robotic methods, while also highlighting the critical role of enforcing physical constraints for safe and robust execution.

FasterDiT: Towards Faster Diffusion Transformers Training without Architecture Modification

Diffusion Transformers (DiT) have attracted significant attention in research. However, they suffer from a slow convergence rate. In this paper, we aim to accelerate DiT training without any architectural modification. We identify the following issues in the training process: firstly, certain training strategies do not consistently perform well across different data. Secondly, the effectiveness of supervision at specific timesteps is limited. In response, we propose the following contributions: (1) We introduce a new perspective for interpreting the failure of the strategies. Specifically, we slightly extend the definition of Signal-to-Noise Ratio (SNR) and suggest observing the Probability Density Function (PDF) of SNR to understand the essence of the data robustness of the strategy. (2) We conduct numerous experiments and report over one hundred experimental results to empirically summarize a unified accelerating strategy from the perspective of PDF. (3) We develop a new supervision method that further accelerates the training process of DiT. Based on them, we propose FasterDiT, an exceedingly simple and practicable design strategy. With few lines of code modifications, it achieves 2.30 FID on ImageNet 256 resolution at 1000k iterations, which is comparable to DiT (2.27 FID) but 7 times faster in training.

M${}^{3}$Bench: Benchmarking Whole-body Motion Generation for Mobile Manipulation in 3D Scenes

We propose M^3Bench, a new benchmark for whole-body motion generation for mobile manipulation tasks. Given a 3D scene context, M^3Bench requires an embodied agent to understand its configuration, environmental constraints and task objectives, then generate coordinated whole-body motion trajectories for object rearrangement tasks. M^3Bench features 30k object rearrangement tasks across 119 diverse scenes, providing expert demonstrations generated by our newly developed M^3BenchMaker. This automatic data generation tool produces coordinated whole-body motion trajectories from high-level task instructions, requiring only basic scene and robot information. Our benchmark incorporates various task splits to assess generalization across different dimensions and leverages realistic physics simulation for trajectory evaluation. Through extensive experimental analyses, we reveal that state-of-the-art models still struggle with coordinated base-arm motion while adhering to environment-context and task-specific constraints, highlighting the need to develop new models that address this gap. Through M^3Bench, we aim to facilitate future robotics research towards more adaptive and capable mobile manipulation in diverse, real-world environments.

Mamba Capsule Routing Towards Part-Whole Relational Camouflaged Object Detection

The part-whole relational property endowed by Capsule Networks (CapsNets) has been known successful for camouflaged object detection due to its segmentation integrity. However, the previous Expectation Maximization (EM) capsule routing algorithm with heavy computation and large parameters obstructs this trend. The primary attribution behind lies in the pixel-level capsule routing. Alternatively, in this paper, we propose a novel mamba capsule routing at the type level. Specifically, we first extract the implicit latent state in mamba as capsule vectors, which abstract type-level capsules from pixel-level versions. These type-level mamba capsules are fed into the EM routing algorithm to get the high-layer mamba capsules, which greatly reduce the computation and parameters caused by the pixel-level capsule routing for part-whole relationships exploration. On top of that, to retrieve the pixel-level capsule features for further camouflaged prediction, we achieve this on the basis of the low-layer pixel-level capsules with the guidance of the correlations from adjacent-layer type-level mamba capsules. Extensive experiments on three widely used COD benchmark datasets demonstrate that our method significantly outperforms state-of-the-arts. Code has been available on https://github.com/Liangbo-Cheng/mamba\_capsule.

ControlAR: Controllable Image Generation with Autoregressive Models

Autoregressive (AR) models have reformulated image generation as next-token prediction, demonstrating remarkable potential and emerging as strong competitors to diffusion models. However, control-to-image generation, akin to ControlNet, remains largely unexplored within AR models. Although a natural approach, inspired by advancements in Large Language Models, is to tokenize control images into tokens and prefill them into the autoregressive model before decoding image tokens, it still falls short in generation quality compared to ControlNet and suffers from inefficiency. To this end, we introduce ControlAR, an efficient and effective framework for integrating spatial controls into autoregressive image generation models. Firstly, we explore control encoding for AR models and propose a lightweight control encoder to transform spatial inputs (e.g., canny edges or depth maps) into control tokens. Then ControlAR exploits the conditional decoding method to generate the next image token conditioned on the per-token fusion between control and image tokens, similar to positional encodings. Compared to prefilling tokens, using conditional decoding significantly strengthens the control capability of AR models but also maintains the model's efficiency. Furthermore, the proposed ControlAR surprisingly empowers AR models with arbitrary-resolution image generation via conditional decoding and specific controls. Extensive experiments can demonstrate the controllability of the proposed ControlAR for the autoregressive control-to-image generation across diverse inputs, including edges, depths, and segmentation masks. Furthermore, both quantitative and qualitative results indicate that ControlAR surpasses previous state-of-the-art controllable diffusion models, e.g., ControlNet++. Code, models, and demo will soon be available at https://github.com/hustvl/ControlAR.

Dynamic 2D Gaussians: Geometrically accurate radiance fields for dynamic objects

Reconstructing objects and extracting high-quality surfaces play a vital role in the real world. Current 4D representations show the ability to render high-quality novel views for dynamic objects but cannot reconstruct high-quality meshes due to their implicit or geometrically inaccurate representations. In this paper, we propose a novel representation that can reconstruct accurate meshes from sparse image input, named Dynamic 2D Gaussians (D-2DGS). We adopt 2D Gaussians for basic geometry representation and use sparse-controlled points to capture 2D Gaussian's deformation. By extracting the object mask from the rendered high-quality image and masking the rendered depth map, a high-quality dynamic mesh sequence of the object can be extracted. Experiments demonstrate that our D-2DGS is outstanding in reconstructing high-quality meshes from sparse input. More demos and code are available at https://github.com/hustvl/Dynamic-2DGS.

PersonViT: Large-scale Self-supervised Vision Transformer for Person Re-Identificat

Person Re-Identification (ReID) aims to retrieve relevant individuals in non-overlapping camera images and has a wide range of applications in the field of public safety. In recent years, with the development of Vision Transformer (ViT) and self-supervised learning techniques, the performance of person ReID based on self-supervised pre-training has been greatly improved. Person ReID requires extracting highly discriminative local fine-grained features of the human body, while traditional ViT is good at extracting context-related global features, making it difficult to focus on local human body features. To this end, this article introduces the recently emerged Masked Image Modeling (MIM) self-supervised learning method into person ReID, and effectively extracts high-quality global and local features through large-scale unsupervised pre-training by combining masked image modeling and discriminative contrastive learning, and then conducts supervised fine-tuning training in the person ReID task. This person feature extraction method based on ViT with masked image modeling (PersonViT) has the good characteristics of unsupervised, scalable, and strong generalization capabilities, overcoming the problem of difficult annotation in supervised person ReID, and achieves state-of-the-art results on publicly available benchmark datasets, including MSMT17, Market1501, DukeMTMC-reID, and Occluded-Duke. The code and pre-trained models of the PersonViT method are released at https://github.com/hustvl/PersonViT to promote further research in the person ReID fie

XS-VID: An Extremely Small Video Object Detection Dataset

Small Video Object Detection (SVOD) is a crucial subfield in modern computer vision, essential for early object discovery and detection. However, existing SVOD datasets are scarce and suffer from issues such as insufficiently small objects, limited object categories, and lack of scene diversity, leading to unitary application scenarios for corresponding methods. To address this gap, we develop the XS-VID dataset, which comprises aerial data from various periods and scenes, and annotates eight major object categories. To further evaluate existing methods for detecting extremely small objects, XS-VID extensively collects three types of objects with smaller pixel areas: extremely small (\textit{es}, $0\sim12^2$), relatively small (\textit{rs}, $12^2\sim20^2$), and generally small (\textit{gs}, $20^2\sim32^2$). XS-VID offers unprecedented breadth and depth in covering and quantifying minuscule objects, significantly enriching the scene and object diversity in the dataset. Extensive validations on XS-VID and the publicly available VisDrone2019VID dataset show that existing methods struggle with small object detection and significantly underperform compared to general object detectors. Leveraging the strengths of previous methods and addressing their weaknesses, we propose YOLOFT, which enhances local feature associations and integrates temporal motion features, significantly improving the accuracy and stability of SVOD. Our datasets and benchmarks are available at \url{https://gjhhust.github.io/XS-VID/}.