Penghao Wu

Li Chen, Penghao Wu, Kashyap Chitta et al. · pku

The autonomous driving community has witnessed a rapid growth in approaches that embrace an end-to-end algorithm framework, utilizing raw sensor input to generate vehicle motion plans, instead of concentrating on individual tasks such as detection and motion prediction. End-to-end systems, in comparison to modular pipelines, benefit from joint feature optimization for perception and planning. This field has flourished due to the availability of large-scale datasets, closed-loop evaluation, and the increasing need for autonomous driving algorithms to perform effectively in challenging scenarios. In this survey, we provide a comprehensive analysis of more than 270 papers, covering the motivation, roadmap, methodology, challenges, and future trends in end-to-end autonomous driving. We delve into several critical challenges, including multi-modality, interpretability, causal confusion, robustness, and world models, amongst others. Additionally, we discuss current advancements in foundation models and visual pre-training, as well as how to incorporate these techniques within the end-to-end driving framework. we maintain an active repository that contains up-to-date literature and open-source projects at https://github.com/OpenDriveLab/End-to-end-Autonomous-Driving.

Shengchao Hu, Li Chen, Penghao Wu et al. · pku

Many existing autonomous driving paradigms involve a multi-stage discrete pipeline of tasks. To better predict the control signals and enhance user safety, an end-to-end approach that benefits from joint spatial-temporal feature learning is desirable. While there are some pioneering works on LiDAR-based input or implicit design, in this paper we formulate the problem in an interpretable vision-based setting. In particular, we propose a spatial-temporal feature learning scheme towards a set of more representative features for perception, prediction and planning tasks simultaneously, which is called ST-P3. Specifically, an egocentric-aligned accumulation technique is proposed to preserve geometry information in 3D space before the bird's eye view transformation for perception; a dual pathway modeling is devised to take past motion variations into account for future prediction; a temporal-based refinement unit is introduced to compensate for recognizing vision-based elements for planning. To the best of our knowledge, we are the first to systematically investigate each part of an interpretable end-to-end vision-based autonomous driving system. We benchmark our approach against previous state-of-the-arts on both open-loop nuScenes dataset as well as closed-loop CARLA simulation. The results show the effectiveness of our method. Source code, model and protocol details are made publicly available at https://github.com/OpenPerceptionX/ST-P3.

Penghao Wu, Xiaosong Jia, Li Chen et al. · pku

Current end-to-end autonomous driving methods either run a controller based on a planned trajectory or perform control prediction directly, which have spanned two separately studied lines of research. Seeing their potential mutual benefits to each other, this paper takes the initiative to explore the combination of these two well-developed worlds. Specifically, our integrated approach has two branches for trajectory planning and direct control, respectively. The trajectory branch predicts the future trajectory, while the control branch involves a novel multi-step prediction scheme such that the relationship between current actions and future states can be reasoned. The two branches are connected so that the control branch receives corresponding guidance from the trajectory branch at each time step. The outputs from two branches are then fused to achieve complementary advantages. Our results are evaluated in the closed-loop urban driving setting with challenging scenarios using the CARLA simulator. Even with a monocular camera input, the proposed approach ranks first on the official CARLA Leaderboard, outperforming other complex candidates with multiple sensors or fusion mechanisms by a large margin. The source code is publicly available at https://github.com/OpenPerceptionX/TCP

Li Chen, Tutian Tang, Zhitian Cai et al. · pku

Equipped with a wide span of sensors, predominant autonomous driving solutions are becoming more modular-oriented for safe system design. Though these sensors have laid a solid foundation, most massive-production solutions up to date still fall into L2 phase. Among these, Comma.ai comes to our sight, claiming one $999 aftermarket device mounted with a single camera and board inside owns the ability to handle L2 scenarios. Together with open-sourced software of the entire system released by Comma.ai, the project is named Openpilot. Is it possible? If so, how is it made possible? With curiosity in mind, we deep-dive into Openpilot and conclude that its key to success is the end-to-end system design instead of a conventional modular framework. The model is briefed as Supercombo, and it can predict the ego vehicle's future trajectory and other road semantics on the fly from monocular input. Unfortunately, the training process and massive amount of data to make all these work are not publicly available. To achieve an intensive investigation, we try to reimplement the training details and test the pipeline on public benchmarks. The refactored network proposed in this work is referred to as OP-Deepdive. For a fair comparison of our version to the original Supercombo, we introduce a dual-model deployment scheme to test the driving performance in the real world. Experimental results on nuScenes, Comma2k19, CARLA, and in-house realistic scenarios verify that a low-cost device can indeed achieve most L2 functionalities and be on par with the original Supercombo model. In this report, we would like to share our latest findings, shed some light on the new perspective of end-to-end autonomous driving from an industrial product-level side, and potentially inspire the community to continue improving the performance. Our code, benchmarks are at https://github.com/OpenPerceptionX/Openpilot-Deepdive.

Xiaosong Jia, Penghao Wu, Li Chen et al. · pku

Encoding a driving scene into vector representations has been an essential task for autonomous driving that can benefit downstream tasks e.g. trajectory prediction. The driving scene often involves heterogeneous elements such as the different types of objects (agents, lanes, traffic signs) and the semantic relations between objects are rich and diverse. Meanwhile, there also exist relativity across elements, which means that the spatial relation is a relative concept and need be encoded in a ego-centric manner instead of in a global coordinate system. Based on these observations, we propose Heterogeneous Driving Graph Transformer (HDGT), a backbone modelling the driving scene as a heterogeneous graph with different types of nodes and edges. For heterogeneous graph construction, we connect different types of nodes according to diverse semantic relations. For spatial relation encoding, the coordinates of the node as well as its in-edges are in the local node-centric coordinate system. For the aggregation module in the graph neural network (GNN), we adopt the transformer structure in a hierarchical way to fit the heterogeneous nature of inputs. Experimental results show that HDGT achieves state-of-the-art performance for the task of trajectory prediction, on INTERACTION Prediction Challenge and Waymo Open Motion Challenge.

Penghao Wu, Li Chen, Hongyang Li et al. · pku

Witnessing the impressive achievements of pre-training techniques on large-scale data in the field of computer vision and natural language processing, we wonder whether this idea could be adapted in a grab-and-go spirit, and mitigate the sample inefficiency problem for visuomotor driving. Given the highly dynamic and variant nature of the input, the visuomotor driving task inherently lacks view and translation invariance, and the visual input contains massive irrelevant information for decision making, resulting in predominant pre-training approaches from general vision less suitable for the autonomous driving task. To this end, we propose PPGeo (Policy Pre-training via Geometric modeling), an intuitive and straightforward fully self-supervised framework curated for the policy pretraining in visuomotor driving. We aim at learning policy representations as a powerful abstraction by modeling 3D geometric scenes on large-scale unlabeled and uncalibrated YouTube driving videos. The proposed PPGeo is performed in two stages to support effective self-supervised training. In the first stage, the geometric modeling framework generates pose and depth predictions simultaneously, with two consecutive frames as input. In the second stage, the visual encoder learns driving policy representation by predicting the future ego-motion and optimizing with the photometric error based on current visual observation only. As such, the pre-trained visual encoder is equipped with rich driving policy related representations and thereby competent for multiple visuomotor driving tasks. Extensive experiments covering a wide span of challenging scenarios have demonstrated the superiority of our proposed approach, where improvements range from 2% to even over 100% with very limited data.

Haiwen Diao, Jiahao Wang, Penghao Wu et al.

Current vision-language models (VLMs) typically stitch together separate image encoders and language decoders via multi-stage alignment, a modular framework that inevitably fragments pixel-level signals across frames and scatters early pixel-word interactions. In parallel, native VLMs, despite impressive performance on single images, remain largely unexplored in multi-image, video understanding, and spatial intelligence. Hence, we introduce NEO-ov, a native foundation model that learns cross-frame and pixel-word correspondence end-to-end, without any external encoders, auxiliary adapters, or post-hoc fusion. By eliminating module boundaries entirely, NEO-ov enables fine-grained and unified spatiotemporal modeling to emerge natively inside the model. Notably, NEO-ov largely narrows the gap to modular counterparts while excelling at fine-grained visual perception, validating that native "one-vision" architectures are not only feasible but competitive at scale. Beyond empirical performance, we unveil systematic architectural analyses and detailed training recipes to facilitate subsequent native multimodal modeling. Our code and models are publicly available at: https://github.com/EvolvingLMMs-Lab/NEO.

Jing Liang, Li Niu, Penghao Wu et al.

Inharmonious region localization aims to localize the region in a synthetic image which is incompatible with surrounding background. The inharmony issue is mainly attributed to the color and illumination inconsistency produced by image editing techniques. In this work, we tend to transform the input image to another color space to magnify the domain discrepancy between inharmonious region and background, so that the model can identify the inharmonious region more easily. To this end, we present a novel framework consisting of a color mapping module and an inharmonious region localization network, in which the former is equipped with a novel domain discrepancy magnification loss and the latter could be an arbitrary localization network. Extensive experiments on image harmonization dataset show the superiority of our designed framework. Our code is available at https://github.com/bcmi/MadisNet-Inharmonious-Region-Localization.

Penghao Wu, Li Niu, Liqing Zhang

With the prevalence of image editing techniques, users can create fantastic synthetic images, but the image quality may be compromised by the color/illumination discrepancy between the manipulated region and background. Inharmonious region localization aims to localize the inharmonious region in a synthetic image. In this work, we attempt to leverage auxiliary style feature to facilitate this task. Specifically, we propose a novel color mapping module and a style feature loss to extract discriminative style features containing task-relevant color/illumination information. Based on the extracted style features, we also propose a novel style voting module to guide the localization of inharmonious region. Moreover, we introduce semantic information into the style voting module to achieve further improvement. Our method surpasses the existing methods by a large margin on the benchmark dataset.

Penghao Wu, Saining Xie

When we look around and perform complex tasks, how we see and selectively process what we see is crucial. However, the lack of this visual search mechanism in current multimodal LLMs (MLLMs) hinders their ability to focus on important visual details, especially when handling high-resolution and visually crowded images. To address this, we introduce V*, an LLM-guided visual search mechanism that employs the world knowledge in LLMs for efficient visual querying. When combined with an MLLM, this mechanism enhances collaborative reasoning, contextual understanding, and precise targeting of specific visual elements. This integration results in a new MLLM meta-architecture, named Show, sEArch, and TelL (SEAL). We further create V*Bench, a benchmark specifically designed to evaluate MLLMs in their ability to process high-resolution images and focus on visual details. Our study highlights the necessity of incorporating visual search capabilities into multimodal systems. The code is available https://github.com/penghao-wu/vstar.

Penghao Wu, Li Niu, Jing Liang et al.

Synthetic images created by image editing operations are prevalent, but the color or illumination inconsistency between the manipulated region and background may make it unrealistic. Thus, it is important yet challenging to localize the inharmonious region to improve the quality of synthetic image. Inspired by the classic clustering algorithm, we aim to group pixels into two clusters: inharmonious cluster and background cluster by inserting a novel Recurrent Self-Reasoning (RSR) module into the bottleneck of UNet structure. The mask output from RSR module is provided for the decoder as attention guidance. Finally, we adaptively combine the masks from RSR and the decoder to form our final mask. Experimental results on the image harmonization dataset demonstrate that our method achieves competitive performance both quantitatively and qualitatively.

Haiwen Diao, Penghao Wu, Hanming Deng et al.

Recent large vision-language models (VLMs) remain fundamentally constrained by a persistent dichotomy: understanding and generation are treated as distinct problems, leading to fragmented architectures, cascaded pipelines, and misaligned representation spaces. We argue that this divide is not merely an engineering artifact, but a structural limitation that hinders the emergence of native multimodal intelligence. Hence, we introduce SenseNova-U1, a native unified multimodal paradigm built upon NEO-unify, in which understanding and generation evolve as synergistic views of a single underlying process. We launch two native unified variants, SenseNova-U1-8B-MoT and SenseNova-U1-A3B-MoT, built on dense (8B) and mixture-of-experts (30B-A3B) understanding baselines, respectively. Designed from first principles, they rival top-tier understanding-only VLMs across text understanding, vision-language perception, knowledge reasoning, agentic decision-making, and spatial intelligence. Meanwhile, they deliver strong semantic consistency and visual fidelity, excelling in conventional or knowledge-intensive any-to-image (X2I) synthesis, complex text-rich infographic generation, and interleaved vision-language generation, with or without think patterns. Beyond performance, we show detailed model design, data preprocessing, pre-/post-training, and inference strategies to support community research. Last but not least, preliminary evidence demonstrates that our models extend beyond perception and generation, performing strongly in vision-language-action (VLA) and world model (WM) scenarios. This points toward a broader roadmap where models do not translate between modalities, but think and act across them in a native manner. Multimodal AI is no longer about connecting separate systems, but about building a unified one and trusting the necessary capabilities to emerge from within.

Penghao Wu, Lewei Lu, Ziwei Liu

Large multimodal models excel in multimodal tasks but face significant computational challenges due to excessive computation on visual tokens. Unlike token reduction methods that focus on token-level redundancy, we identify and study the computation-level redundancy on vision tokens to ensure no information loss. Our key insight is that vision tokens from the pretrained vision encoder do not necessarily require all the heavy operations (e.g., self-attention, FFNs) in decoder-only LMMs and could be processed more lightly with proper designs. We designed a series of experiments to discover and progressively squeeze out the vision-related computation redundancy. Based on our findings, we propose ProxyV, a novel approach that utilizes proxy vision tokens to alleviate the computational burden on original vision tokens. ProxyV enhances efficiency without compromising performance and can even yield notable performance gains in scenarios with more moderate efficiency improvements. Furthermore, the flexibility of ProxyV is demonstrated through its combination with token reduction methods to boost efficiency further. The code will be made public at this https://github.com/penghao-wu/ProxyV URL.

Kairui Hu, Penghao Wu, Fanyi Pu et al.

Humans acquire knowledge through three cognitive stages: perceiving information, comprehending knowledge, and adapting knowledge to solve novel problems. Videos serve as an effective medium for this learning process, facilitating a progression through these cognitive stages. However, existing video benchmarks fail to systematically evaluate the knowledge acquisition capabilities in Large Multimodal Models (LMMs). To address this gap, we introduce Video-MMMU, a multi-modal, multi-disciplinary benchmark designed to assess LMMs' ability to acquire and utilize knowledge from videos. Video-MMMU features a curated collection of 300 expert-level videos and 900 human-annotated questions across six disciplines, evaluating knowledge acquisition through stage-aligned question-answer pairs: Perception, Comprehension, and Adaptation. A proposed knowledge gain metric, Δknowledge, quantifies improvement in performance after video viewing. Evaluation of LMMs reveals a steep decline in performance as cognitive demands increase and highlights a significant gap between human and model knowledge acquisition, underscoring the need for methods to enhance LMMs' capability to learn and adapt from videos.

Shulin Tian, Ruiqi Wang, Hongming Guo et al.

We introduce Ego-R1, a novel framework for reasoning over ultra-long (i.e., in days and weeks) egocentric videos, which leverages a structured Chain-of-Tool-Thought (CoTT) process, orchestrated by an Ego-R1 Agent trained via reinforcement learning (RL). Inspired by human problem-solving strategies, CoTT decomposes complex reasoning into modular steps, with the RL agent invoking specific tools, one per step, to iteratively and collaboratively answer sub-questions tackling such tasks as temporal retrieval and multi-modal understanding. We design a two-stage training paradigm involving supervised finetuning (SFT) of a pretrained language model using CoTT data and RL to enable our agent to dynamically propose step-by-step tools for long-range reasoning. To facilitate training, we construct a dataset called Ego-R1 Data, which consists of Ego-CoTT-25K for SFT and Ego-QA-4.4K for RL. Furthermore, our Ego-R1 agent is evaluated on a newly curated week-long video QA benchmark, Ego-R1 Bench, which contains human-verified QA pairs from hybrid sources. Extensive results demonstrate that the dynamic, tool-augmented chain-of-thought reasoning by our Ego-R1 Agent can effectively tackle the unique challenges of understanding ultra-long egocentric videos, significantly extending the time coverage from few hours to a week.

Penghao Wu, Shengnan Ma, Bo Wang et al.

Multimodal Large Language Models (MLLMs) have shown great potential in revolutionizing Graphical User Interface (GUI) automation. However, existing GUI models mostly rely on learning from nearly error-free offline trajectories, thus lacking reflection and error recovery capabilities. To bridge this gap, we propose GUI-Reflection, a novel framework that explicitly integrates self-reflection and error correction capabilities into end-to-end multimodal GUI models throughout dedicated training stages: GUI-specific pre-training, offline supervised fine-tuning (SFT), and online reflection tuning. GUI-reflection enables self-reflection behavior emergence with fully automated data generation and learning processes without requiring any human annotation. Specifically, 1) we first propose scalable data pipelines to automatically construct reflection and error correction data from existing successful trajectories. While existing GUI models mainly focus on grounding and UI understanding ability, we propose the GUI-Reflection Task Suite to learn and evaluate reflection-oriented abilities explicitly. 2) Furthermore, we built a diverse and efficient environment for online training and data collection of GUI models on mobile devices. 3) We also present an iterative online reflection tuning algorithm leveraging the proposed environment, enabling the model to continuously enhance its reflection and error correction abilities. Our framework equips GUI agents with self-reflection and correction capabilities, paving the way for more robust, adaptable, and intelligent GUI automation, with all data, models, environments, and tools to be released publicly.

Penghao Wu, Yushan Zhang, Haiwen Diao et al.

Reinforcement learning based post-training has recently emerged as a powerful paradigm for enhancing the alignment and reasoning capabilities of multimodal large language models (MLLMs). While vision-centric post-training is crucial for enhancing MLLMs' intrinsic understanding of visual signals, current post-training paradigms are predominantly text-centric, where dense visual inputs are only leveraged to extract sparse cues for text-based reasoning. There exist a few approaches in this direction, however, they often still rely on text as an intermediate mediator or introduce additional visual generative designs. In this work, we introduce Visual Jigsaw, a generic self-supervised post-training framework designed to strengthen visual understanding in MLLMs. Visual Jigsaw is formulated as a general ordering task: visual inputs are partitioned, shuffled, and the model must reconstruct the visual information by producing the correct permutation in natural language. This naturally aligns with reinforcement learning from verifiable rewards (RLVR), requires no additional visual generative components, and derives its supervisory signal automatically without any annotations. We instantiate Visual Jigsaw across three visual modalities, including images, videos, and 3D data. Extensive experiments demonstrate substantial improvements in fine-grained perception, temporal reasoning, and 3D spatial understanding. Our findings highlight the potential of self-supervised vision-centric tasks in post-training MLLMs and aim to inspire further research on vision-centric pretext designs. Project Page: https://penghao-wu.github.io/visual_jigsaw/

Shengbang Tong, Ellis Brown, Penghao Wu et al.

We introduce Cambrian-1, a family of multimodal LLMs (MLLMs) designed with a vision-centric approach. While stronger language models can enhance multimodal capabilities, the design choices for vision components are often insufficiently explored and disconnected from visual representation learning research. This gap hinders accurate sensory grounding in real-world scenarios. Our study uses LLMs and visual instruction tuning as an interface to evaluate various visual representations, offering new insights into different models and architectures -- self-supervised, strongly supervised, or combinations thereof -- based on experiments with over 20 vision encoders. We critically examine existing MLLM benchmarks, address the difficulties involved in consolidating and interpreting results from various tasks, and introduce a new vision-centric benchmark, CV-Bench. To further improve visual grounding, we propose the Spatial Vision Aggregator (SVA), a dynamic and spatially-aware connector that integrates high-resolution vision features with LLMs while reducing the number of tokens. Additionally, we discuss the curation of high-quality visual instruction-tuning data from publicly available sources, emphasizing the importance of data source balancing and distribution ratio. Collectively, Cambrian-1 not only achieves state-of-the-art performance but also serves as a comprehensive, open cookbook for instruction-tuned MLLMs. We provide model weights, code, supporting tools, datasets, and detailed instruction-tuning and evaluation recipes. We hope our release will inspire and accelerate advancements in multimodal systems and visual representation learning.

Jiazhi Yang, Shenyuan Gao, Yihang Qiu et al.

In this paper, we introduce the first large-scale video prediction model in the autonomous driving discipline. To eliminate the restriction of high-cost data collection and empower the generalization ability of our model, we acquire massive data from the web and pair it with diverse and high-quality text descriptions. The resultant dataset accumulates over 2000 hours of driving videos, spanning areas all over the world with diverse weather conditions and traffic scenarios. Inheriting the merits from recent latent diffusion models, our model, dubbed GenAD, handles the challenging dynamics in driving scenes with novel temporal reasoning blocks. We showcase that it can generalize to various unseen driving datasets in a zero-shot manner, surpassing general or driving-specific video prediction counterparts. Furthermore, GenAD can be adapted into an action-conditioned prediction model or a motion planner, holding great potential for real-world driving applications.

Xiaosong Jia, Penghao Wu, Li Chen et al.

End-to-end autonomous driving has made impressive progress in recent years. Existing methods usually adopt the decoupled encoder-decoder paradigm, where the encoder extracts hidden features from raw sensor data, and the decoder outputs the ego-vehicle's future trajectories or actions. Under such a paradigm, the encoder does not have access to the intended behavior of the ego agent, leaving the burden of finding out safety-critical regions from the massive receptive field and inferring about future situations to the decoder. Even worse, the decoder is usually composed of several simple multi-layer perceptrons (MLP) or GRUs while the encoder is delicately designed (e.g., a combination of heavy ResNets or Transformer). Such an imbalanced resource-task division hampers the learning process. In this work, we aim to alleviate the aforementioned problem by two principles: (1) fully utilizing the capacity of the encoder; (2) increasing the capacity of the decoder. Concretely, we first predict a coarse-grained future position and action based on the encoder features. Then, conditioned on the position and action, the future scene is imagined to check the ramification if we drive accordingly. We also retrieve the encoder features around the predicted coordinate to obtain fine-grained information about the safety-critical region. Finally, based on the predicted future and the retrieved salient feature, we refine the coarse-grained position and action by predicting its offset from ground-truth. The above refinement module could be stacked in a cascaded fashion, which extends the capacity of the decoder with spatial-temporal prior knowledge about the conditioned future. We conduct experiments on the CARLA simulator and achieve state-of-the-art performance in closed-loop benchmarks. Extensive ablation studies demonstrate the effectiveness of each proposed module.