[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-463448-105":59,"doc-detail-463448-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","hipro-ad-sparse-trajectory-transformer-for-end-to-end-autonomous-driving-with-hybrid-spatiotemporal-attention","HiPro-AD - Sparse Trajectory Transformer for End-to-End Autonomous Driving with Hybrid Spatiotemporal Attention","","End-to-end (E2E) autonomous driving maps raw sensor data directly to vehicle controls, reducing error propagation common in modular pipelines. Dense Bird’s-Eye-View (BEV) approaches, however, bring high computation and require complex trajectory postprocessing. HiPro-AD proposes a sparse, proposal-centric E2E planning framework that departs from dense BEV paradigms by using an IM-ResNet-34 encoder and an STFormer transformer for efficient multi-view fusion and temporal context. Sparse computation targets proposal-relevant regions, and trajectory selection uses a pairwise ranking scorer.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/hipro-ad-sparse-trajectory-transformer-for-end-to-end-autonomous-driving-with-hybrid-spatiotemporal-attention/463448/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/hipro-ad-sparse-trajectory-transformer-for-end-to-end-autonomous-driving-with-hybrid-spatiotemporal-attention/463448.png","ImageObject",300,407,{"name":92,"@type":93},"A glass of water","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-15","2026-09-30",true,{"@type":102,"interactionType":103,"userInteractionCount":52},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"What problem does HiPro-AD target in end-to-end autonomous driving?","Question",{"text":112,"@type":113},"It targets the computational burden and postprocessing complexity of dense BEV-based E2E methods, which rely heavily on dense feature maps for trajectory generation.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"How does HiPro-AD achieve sparse efficiency?",{"text":117,"@type":113},"HiPro-AD uses a proposal-centric mechanism so the transformer fuses multi-view spatial features with historical temporal context while restricting computation to regions relevant to sparse trajectory proposals.",{"name":119,"@type":110,"acceptedAnswer":120},"How are trajectories selected in HiPro-AD?",{"text":121,"@type":113},"A Pairwise Ranking Scorer refines trajectory selection by identifying the best plan from multiple candidates based on relative quality.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},463448,1791160383,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":52,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":144,"read_time":145,"preview_page_count":19,"preview_urls":146},962090760832,"https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d","Article  \nHiPro-AD: Sparse Trajectory Transformer for End-to-End Autonomous Driving with Hybrid Spatiotemporal Attention  \nBing Chen 1,†, Gaopeng Wang 2,3,†, Jiandong Yang 4, Shaoliang Huang 4, Xinhe Qian 1, Bin Huang 3, * and Guanlun Guo 2,3, *  \nAcademic Editors: Araceli Sanchis de Miguel and Agapito Ledezma Espino  \nReceived: 26 November 2025  \nRevised: 20 December 2025  \nAccepted: 24 December 2025  \nPublished: 26 December 2025  \nCopyright: © 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license.  \n1 Shandong Electric Power Engineering Consulting Institute Corp., Ltd., Jinan 250100, China; [chenbing@spic.com.cn](chenbing@spic.com.cn) (B.C.); [qianxinhe@spic.com.cn](qianxinhe@spic.com.cn) (X.Q.)  \n2 Hubei Longzhong Laboratory, Xiangyang 441106, China; [whut1749981935@163.com](whut1749981935@163.com)  \n3 School of Automotive Engineering, Wuhan University of Technology, Wuhan 430070, China  \n4 Sdic Qinzhou Second Electric Power Co., Ltd., Qinzhou 535000, China; [10022685@sdic.com.cn](10022685@sdic.com.cn) (J.Y.); [18177969228@189.cn](18177969228@189.cn) (S.H.)  \n* [Correspondence: huangb@whut.edu.cn](Correspondence: huangb@whut.edu.cn) (B.H.); [glguo@whut.edu.cn](glguo@whut.edu.cn) (G.G.)† These authors contributed equally to this work.  \nAbstract  \nEnd-to-end (E2E) autonomous driving offers a promising alternative to traditional modular pipelines by mapping raw sensor data directly to vehicle controls, thereby mitigating error propagation. However, prevalent approaches largely rely on dense Bird’s-Eye-View (BEV) feature maps, which incur high computational overhead and necessitate complex postprocessing for trajectory generation. To address these limitations, we propose HiPro-AD, a proposal-centric sparse E2E planning framework that fundamentally diverges from dense BEV paradigms. HiPro-AD integrates an efficiency-oriented IM-ResNet-34 encoder with a novel STFormer. This transformer dynamically fuses multi-view spatial features and historical temporal context via a proposal-anchored mechanism, focusing computation strictly on regions relevant to sparse trajectory proposals. Furthermore, trajectory selection is refined by a Pairwise Ranking Scorer, which identifies the optimal plan from diverse candidates based on relative quality. On the NAVSIM benchmark, HiPro-AD achievesa PDMS of 92.6 using only camera input, surpassing prior dense BEV and multimodal methods. On the closed-loop Bench2Drive benchmark, it attains a 37.31% success rate and a driving score of 65.48 with a latency of 67 ms, demonstrating real-time capability. These results validate the efficiency and robustness of our sparse paradigm in complex driving scenarios.  \nKeywords: autonomous driving; bird’s eye view; trajectory planning  \n1. Introduction  \nEnd-to-end (E2E) autonomous driving, which directly regresses control commands or future trajectories from raw sensor inputs, has emerged as a promising alternative to traditional modular pipelines. By eliminating hand-crafted intermediate representationsand heuristic rules, E2E systems mitigate error propagation and significantly improve adaptability in complex urban scenarios [1–3] . Meanwhile, robust control methods have tackled safety challenges, such as secure observer-based control under non-Gaussian noises [4] and cloud-based collision avoidance with token bucket shapers [5] . Recent advances, such as UniAD [6], VAD [7], and DiffusionDrive [8], have achieved impressive closed-loop performance by adopting dense Bird’s-Eye-View (BEV) feature grids as the  \nunified scene representation. These dense BEV encoders enable rich spatial reasoning and seamless integration of multi-view imagery through learned view transformation.  \nIn recent years, the research landscape has evolved into several distinct learning paradigms. As summarized in Table 1, these generally include Imitation Le","cbCaiv9Gi1gyejBF","https://ap.wps.com/l/cbCaiv9Gi1gyejBF","pdf",4261958,22,"English","# Introduction\n## Learning paradigms in autonomous driving\n# Related work and comparisons\n# HiPro-AD framework overview\n## Encoder and STFormer\n## Pairwise ranking trajectory selection\n# Experiments and benchmarks\n## NAVSIM results\n## Bench2Drive closed-loop results","[{\"question\":\"What problem does HiPro-AD target in end-to-end autonomous driving?\",\"answer\":\"It targets the computational burden and postprocessing complexity of dense BEV-based E2E methods, which rely heavily on dense feature maps for trajectory generation.\"},{\"question\":\"How does HiPro-AD achieve sparse efficiency?\",\"answer\":\"HiPro-AD uses a proposal-centric mechanism so the transformer fuses multi-view spatial features with historical temporal context while restricting computation to regions relevant to sparse trajectory proposals.\"},{\"question\":\"How are trajectories selected in HiPro-AD?\",\"answer\":\"A Pairwise Ranking Scorer refines trajectory selection by identifying the best plan from multiple candidates based on relative quality.\"}]","HiPro-AD - Sparse Trajectory Transformer for End-to-End Autonomous Driving with Hybrid Spatiotemporal Attention | PDF",1790766596,55,[147,151,155,159],{"page_no":8,"url":148,"width":149,"height":150},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/813c2b657c03fa3f8ff3f07ce80bbbd7.webp",983,1390,{"page_no":14,"url":152,"width":153,"height":154},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/9dfd587f8aa6dcbb937c4c4f32341901.webp",982,1388,{"page_no":81,"url":156,"width":157,"height":158},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/5760506ee256d40cfa884ce2a04c8378.webp",981,1387,{"page_no":19,"url":160,"width":161,"height":162},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/c4f4f671847c54656a3b27a399051d71.webp",980,1385]