论文库
论文库按学习档位分段:精读与中文笔记优先;英文自动卡单独列出,不与中文笔记混计。
公开文献 388 · 精读 23 · 中文笔记 236 · 自动卡 129 · 仅元数据 0。
正式论文池 380 篇 · 基础必读选题 114(可学习 0) · 近期 190(可学习 0) · 观察 283 · 待获取 PDF 304(registry 更新 2026-07-20)。
发现候选池:Topic 38 · 候选 4954 · 含 Asta 证据 31 · 今日刷新 0。
筛选提示:侧栏进入 Topic 查看该域 FRW 论文池;可用站内搜索索引中文笔记正文。
精读与中文笔记
Section titled “精读与中文笔记”AdaWorld: Learning Adaptable World Models with Latent ActionsAdaWorld通过信息瓶颈自监督提取上下文不变的潜在动作,并以此条件预训练自回归世界模型,实现无需训练的动作迁移与高效适应。 · 文献 · 精读 · 2026-07-20
BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal TransformersBEVFormer通过预定义网格BEV查询与时空注意力,从多相机图像与历史BEV中自适应聚合信息,生成支持多感知任务的统一BEV特征。 · 文献 · 精读 · 2026-07-20
DRL-Based Trajectory Tracking for Motion-Related Modules in Autonomous Driving用DRL(结合域随机化与后优化)实现更鲁棒、精确的轨迹跟踪,在高初始速度下显著优于启发式方法。 · 文献 · 精读 · 2026-07-20
DeepFusion: Lidar-Camera Deep Fusion for Multi-Modal 3D Object DetectionDeepFusion在深度特征层融合激光雷达与相机,并用InverseAug与LearnableAlign解决增强导致的对齐难题,从而提升多模态3D目标检测精度与鲁棒性。 · 文献 · 精读 · 2026-07-20
Deformable DETR: Deformable Transformers for End-to-End Object DetectionDeformable DETR用只关注参考点周围少量采样点的可变形(多尺度)注意力替换标准Transformer注意力,使DETR在COCO上以约10倍更少训练epochs取得更好性能(尤其小目标)。 · 文献 · 精读 · 2026-07-20
End-to-End Object Detection with TransformersDETR通过Transformer编码器-解码器与二分图匹配集合损失,将目标检测变为直接并行输出固定大小物体集合的端到端问题,在COCO上达到与优化后Faster R-CNN相当的性能。 · 文献 · 精读 · 2026-07-20
Exploring the Limitations of Behavior Cloning for Autonomous DrivingMetadata-only:Exploring the Limitations of Behavior Cloning for Autonomous Driving · 文献 · 精读 · 2026-07-20
FB-OCC: 3D Occupancy Prediction based on Forward-Backward View TransformationFB-OCC基于前向-后向视图变换,结合联合深度-语义预训练、体素-BEV联合表示、模型扩展与后处理,在nuScenes占用预测上取得54.19% mIoU并获挑战赛第一。 · 文献 · 精读 · 2026-07-20
Is Behavior Cloning All You Need? Understanding Horizon in Imitation LearningMetadata-only:Is Behavior Cloning All You Need? Understanding Horizon in Imitation Learning · 文献 · 精读 · 2026-07-20
Motion planning with sequential convex optimization and convex collision checking如何把带碰撞与动力学约束的运动规划写成**序列凸优化(SCP)**可迭代求解的形式。 · 文献 · 精读 · 2026-07-20
OpenVLA: An Open-Source Vision-Language-Action ModelOpenVLA是基于Llama 2与融合DINOv2+SigLIP视觉编码器的7B开源VLA,在970k Open X-Embodiment轨迹上训练,实现通用多机器人操作与高效微调,优于更大闭源模型。 · 文献 · 精读 · 2026-07-20
PETR: Position Embedding Transformation for Multi-View 3D Object DetectionPETR通过3D位置嵌入变换将多视角2D特征转化为3D位置感知特征,实现简单端到端的多视角3D目标检测并取得SOTA。 · 文献 · 精读 · 2026-07-20
PPAD: Iterative Interactions of Prediction and Planning for End-to-end Autonomous DrivingPPAD以自回归方式在每个时间步交错预测(agent)与规划(ego),通过ego-to-agent/map/BEV层次化交互生成更准确的规划轨迹。 · 文献 · 精读 · 2026-07-20
RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic ControlRT-2 通过将机器人动作 token 化并与网络规模 VQA 等任务共同微调 VLM,直接输出低层控制动作,从而把网页知识迁移到机器人闭环控制并获得涌现能力。 · 文献 · 精读 · 2026-07-20
Reinforced Imitative Trajectory Planning for Urban Automated DrivingRITP通过在actor-critic中集成动作采样器与模仿学习监督,实现多步轨迹规划,配合Transformer贝叶斯奖励与混合优化框架,在nuPlan闭环评估中优于同结构基线并接近SOTA。 · 文献 · 精读 · 2026-07-20
Trajectory optimization of connected and autonomous vehicles at a multilane freeway merging areaMetadata-only:Trajectory optimization of connected and autonomous vehicles at a multilane freeway merging area · 文献 · 精读 · 2026-07-20
VAD: Vectorized Scene Representation for Efficient Autonomous DrivingVAD 将驾驶场景建模为全向量化表示(智能体运动向量与地图向量),通过查询交互与实例级规划约束实现高效、安全的端到端轨迹规划。 · 文献 · 精读 · 2026-07-20
World-in-World: World Models in a Closed-Loop WorldWorld-in-World提出统一闭环在线规划与动作API,在四类具身任务上以任务成功率为首要指标评测世界模型,发现视觉质量不等于成功、后训练action-observation数据扩展优于升级预训练生成器、推理时扩展能显著提升闭环性能。 · 文献 · 精读 · 2026-07-20
nuScenes: A Multimodal Dataset for Autonomous DrivingnuScenes是首个全传感器套件、含雷达与属性标注的大规模多模态自动驾驶数据集(1000个20s场景),并定义了基于中心距离的3D检测/跟踪指标与基线。 · 文献 · 精读 · 2026-07-14
Sparse4D v1 文献笔记如何在多相机 3D 检测中,以少量目标相关的 3D anchors 替代 dense BEV 网格,并融合跨帧观测。 · 文献 · 精读 · 2026-07-11
Sparse4D v2 文献笔记如何将前一帧的高置信目标状态直接作为下一帧 decoder 的输入,以稳定地利用时序信息。 · 文献 · 精读 · 2026-07-11
Sparse4D v3 文献笔记如何在 v2 的 recurrent instance memory 上提升检测质量并输出稳定的跨帧目标 ID。 · 文献 · 精读 · 2026-07-11
SparseDrive 文献笔记如何将 Sparse4D 风格的稀疏场景表示从检测与跟踪扩展到端到端自动驾驶规划。 · 文献 · 精读 · 2026-07-11
2nd Place Solution for CVPR2024 E2E Challenge: End-to-End Autonomous Driving Using Vision Language ModelMetadata-only:2nd Place Solution for CVPR2024 E2E Challenge: End-to-End Autonomous Driving Using Vision Language Model · 文献 · 中文笔记 · 2026-07-19
3D Object Detection for Autonomous Driving: A Comprehensive SurveyMetadata-only:3D Object Detection for Autonomous Driving: A Comprehensive Survey · 文献 · 中文笔记 · 2026-07-19
3D Object Detection for Autonomous Driving: A Survey全面综述自动驾驶3D目标检测,从模态分类方法并分析挑战与SOTA,指出领域仍处早期阶段。 · 文献 · 中文笔记 · 2026-07-19
3D Scene Generation: A Survey本文系统综述3D场景生成,将其组织为程序化、神经3D基、图像基与视频基四大范式,并覆盖表示、数据集、评估、应用、挑战与未来方向。 · 文献 · 中文笔记 · 2026-07-19
3D Visual Illusion Depth Estimation3D视觉错觉会误导SOTA单目与双目深度估计,本文构建大规模错觉数据集并提出VLM驱动的单目-双目自适应融合框架,在错觉场景上达到SOTA性能。 · 文献 · 中文笔记 · 2026-07-19
A Communication-Centric 6G-LLM Architecture for Scalable Tactical Autonomous Defense Vehicle Networks用边缘辅助LLM做语义抽象+6G低时延连接,在分层架构下显著降低舰队级TADVN协调时延与通信开销并提升任务成功率(仿真趋势)。 · 文献 · 中文笔记 · 2026-07-19
A Conversational Framework for Human-Robot Collaborative Manipulation with Distributed Generative AI modelsMetadata-only:A Conversational Framework for Human-Robot Collaborative Manipulation with Distributed Generative AI models · 文献 · 中文笔记 · 2026-07-19
A Multi-Drone Multi-View Dataset and Deep Learning Framework for Pedestrian Detection and Tracking提出MATRIX动态多无人机多视角数据集与实时标定+BEV特征融合框架,在复杂遮挡场景下维持约90%检测跟踪精度并优雅应对相机失效。 · 文献 · 中文笔记 · 2026-07-19
A New Literature Review of 3D Object Detection on Autonomous DrivingMetadata-only:A New Literature Review of 3D Object Detection on Autonomous Driving · 文献 · 中文笔记 · 2026-07-19
A Survey of Autonomous Driving: Common Practices and Emerging Technologies本文系统综述自动驾驶常见实践与新兴技术,对比模块化与端到端等架构,指出level 4/5与城市场景鲁棒性仍为开放问题。 · 文献 · 中文笔记 · 2026-07-19
A Survey of Deep Learning-Based Object Detection本文系统综述深度学习目标检测方法,重点分析两阶段与一阶段检测器架构、典型基线、数据集及应用,并讨论高效系统构建与发展趋势。 · 文献 · 中文笔记 · 2026-07-19
A Survey of Large Language Models本文系统综述了大语言模型的背景、关键发现与主流技术(聚焦预训练、适应调优、利用与能力评估),并总结开发资源与未来方向。 · 文献 · 中文笔记 · 2026-07-19
A Survey of Robotic Navigation and Manipulation with Physics Simulators in the Era of Embodied AIMetadata-only:A Survey of Robotic Navigation and Manipulation with Physics Simulators in the Era of Embodied AI · 文献 · 中文笔记 · 2026-07-19
首屏展示 36 / 255 篇;其余请通过侧栏 Topic 或站内搜索进入。
以下多为英文 Paper Card / 自动分析;中文笔记优先读上方区块。需要时再展开对照。
展开英文自动卡(共 117 篇,默认收起)
3D Gaussian Splatting as a New Era: A Survey[自动卡] Metadata-only card for 3D Gaussian Splatting as a New Era: A Survey. · 文献 · 自动卡 · 2026-07-19
A Motion Planning and Tracking Framework for Autonomous Vehicles Based on Artificial Potential Field Elaborated Resistance Network Approach[自动卡] Metadata-only card for A Motion Planning and Tracking Framework for Autonomous Vehicles Based on Artificial Potential Field Elaborated Resistance Network Appro… · 文献 · 自动卡 · 2026-07-19
A Review of Machine Learning and Deep Learning for Object Detection, Semantic Segmentation, and Human Action Recognition in Machine and Robotic Vision[自动卡] Metadata-only card for A Review of Machine Learning and Deep Learning for Object Detection, Semantic Segmentation, and Human Action Recognition in Machine and… · 文献 · 自动卡 · 2026-07-19
A Review of Motion Planning Techniques for Automated Vehicles[自动卡] Metadata-only card for A Review of Motion Planning Techniques for Automated Vehicles. · 文献 · 自动卡 · 2026-07-19
A Review of Sensing Technologies for Indoor Autonomous Mobile Robots[自动卡] Metadata-only card for A Review of Sensing Technologies for Indoor Autonomous Mobile Robots. · 文献 · 自动卡 · 2026-07-19
A Review on Autonomous Vehicles: Progress, Methods and Challenges[自动卡] Metadata-only card for A Review on Autonomous Vehicles: Progress, Methods and Challenges. · 文献 · 自动卡 · 2026-07-19
A Review on Challenges of Autonomous Mobile Robot and Sensor Fusion Methods[自动卡] Metadata-only card for A Review on Challenges of Autonomous Mobile Robot and Sensor Fusion Methods. · 文献 · 自动卡 · 2026-07-19
A Survey of Convolutional Neural Networks: Analysis, Applications, and Prospects[自动卡] Metadata-only card for A Survey of Convolutional Neural Networks: Analysis, Applications, and Prospects. · 文献 · 自动卡 · 2026-07-19
A Survey of Deep RL and IL for Autonomous Driving Policy Learning[自动卡] Metadata-only card for A Survey of Deep RL and IL for Autonomous Driving Policy Learning. · 文献 · 自动卡 · 2026-07-19
A Survey of Deep Reinforcement Learning Algorithms for Motion Planning and Control of Autonomous Vehicles[自动卡] Metadata-only card for A Survey of Deep Reinforcement Learning Algorithms for Motion Planning and Control of Autonomous Vehicles. · 文献 · 自动卡 · 2026-07-19
A Survey of End-to-End Driving: Architectures and Training Methods[自动卡] Metadata-only card for A Survey of End-to-End Driving: Architectures and Training Methods. · 文献 · 自动卡 · 2026-07-19
A Survey of Quantization Methods for Efficient Neural Network Inference[自动卡] Metadata-only card for A Survey of Quantization Methods for Efficient Neural Network Inference. · 文献 · 自动卡 · 2026-07-19
A Survey on Learning-Based Robotic Grasping[自动卡] Metadata-only card for A Survey on Learning-Based Robotic Grasping. · 文献 · 自动卡 · 2026-07-19
A Survey on Multimodal Large Language Models for Autonomous Driving[自动卡] Metadata-only card for A Survey on Multimodal Large Language Models for Autonomous Driving. · 文献 · 自动卡 · 2026-07-19
A Survey on Offline Reinforcement Learning: Taxonomy, Review, and Open Problems[自动卡] Metadata-only card for A Survey on Offline Reinforcement Learning: Taxonomy, Review, and Open Problems. · 文献 · 自动卡 · 2026-07-19
A controllable generative model for generating pavement crack images in complex scenes[自动卡] Metadata-only card for A controllable generative model for generating pavement crack images in complex scenes. · 文献 · 自动卡 · 2026-07-19
A direct yaw moment control frame through model predictive control considering vehicle trajectory tracking performance and handling stability for autonomous driving[自动卡] Metadata-only card for A direct yaw moment control frame through model predictive control considering vehicle trajectory tracking performance and handling stab… · 文献 · 自动卡 · 2026-07-19
A survey of methods for explaining black box models[自动卡] Metadata-only card for A survey of methods for explaining black box models. · 文献 · 自动卡 · 2026-07-19
ARS-DETR: Aspect Ratio-Sensitive Detection Transformer for Aerial Oriented Object Detection[自动卡] Metadata-only card for ARS-DETR: Aspect Ratio-Sensitive Detection Transformer for Aerial Oriented Object Detection. · 文献 · 自动卡 · 2026-07-19
Adversarial Sensor Attack on LiDAR-based Perception in Autonomous Driving[自动卡] Metadata-only card for Adversarial Sensor Attack on LiDAR-based Perception in Autonomous Driving. · 文献 · 自动卡 · 2026-07-19
Agile Autonomous Driving using End-to-End Deep Imitation Learning[自动卡] Metadata-only card for Agile Autonomous Driving using End-to-End Deep Imitation Learning. · 文献 · 自动卡 · 2026-07-19
Anchor DETR: Query Design for Transformer-Based Detector[自动卡] Metadata-only card for Anchor DETR: Query Design for Transformer-Based Detector. · 文献 · 自动卡 · 2026-07-19
Argoverse 2: Next Generation Datasets for Self-Driving Perception and\n Forecasting[自动卡] Metadata-only card for Argoverse 2: Next Generation Datasets for Self-Driving Perception and\n Forecasting. · 文献 · 自动卡 · 2026-07-19
Autonomous Driving Architectures: Insights of Machine Learning and Deep Learning Algorithms[自动卡] Metadata-only card for Autonomous Driving Architectures: Insights of Machine Learning and Deep Learning Algorithms. · 文献 · 自动卡 · 2026-07-19
另有 93 篇自动卡未列出,请通过 Topic 或搜索进入。
暂无。