[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84714-en":3,"doc-seo-84714-105":28,"detail-sidebar-cat-0-en-105":89},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":11,"language":21,"language_code":22,"site_id":23,"html_lang":22,"table_of_contents":24,"faqs":25,"seo_title":13,"seo_description":14,"update_tm":26,"read_time":27},84714,549758252649,"Ivy","https://ap-avatar.wpscdn.com/avatar/8000253669c5317157?_k=1778319167496531819",8,"Research & Report","Neural LiDAR Bundle Adjustment","Recent research delivers strong novel view rendering and 3D scene reconstruction using Neural Radiance Fields (NeRF), with emerging extensions to LiDAR data. However, design differences between RGB NeRFs and LiDAR NeRFs remain underexplored, especially regarding their operating principles. This work shows that the density of volume sampling is pivotal for LiDAR NeRF, and introduces NeLD-BA, a Neural LiDAR Bundle Adjustment method that performs joint optimization of LiDAR map and poses. Experiments on Newer College and FusionPortable datasets validate state-of-the-art multi-view registration and 3D mapping, with code planned for open-source release.","Neural LiDAR Bundle Adjustment  \nChin Yung Anson Hon 1 , Kaicheng Zhang 1 , and Sen Wang 1  \narXiv :2607 .04 169v 1 [ cs .RO] 5 Jul 2026  \nAbstract—Recent research has achieved remarkable novel view rendering and scene reconstruction results with Neural Radiance Field (NeRF), including extensions to the LiDAR modality. Few studies have, however, explored the key design differences between RGB NeRFs and LiDAR NeRFs, particularly considering their underlying working principles. In this work, we provide both theoretical and empirical evidence suggesting that the density of volume sampling plays a significant role in LiDAR NeRF. Based on this finding, we propose a novel Neural LiDAR Bundle Adjustment (NeLD-BA) algorithm, which is tailored using efficient volume sampling of LiDAR rays for joint optimization of LiDAR map and poses. Extensive experiments are performed using the Newer College and FusionPortable datasets to demonstrate the proposed NeLD-BA’s state-of-theart performance in multi-view point cloud registration and 3D mapping. We will open-source our code for the community.  \nI. INTRODUCTION  \nMulti-view LiDAR point cloud registration is a fundamental problem in robotics and computer vision, with applications in autonomous driving, SLAM, and 3D reconstruction. Traditional methods for LiDAR point cloud registration typically rely on feature-based or direct methods, which are sensitive to pose initialization, prone to local minima, and dependent on reliable correspondences. These limitations motivate a shift toward differentiable, optimization-based formulations that refine poses directly from raw geometry without relying on explicit correspondences.  \nRecently, Neural Radiance Fields (NeRFs) [1] have gained significant interest for their powerful capabilities in scene reconstruction and novel view synthesis. A major advancement in RGB image NeRF (RGB NeRF) research is RGB NeRF Bundle Adjustment (BA) [2]–[4], which leverages the differentiability of camera projection to jointly optimize the scene neural representation and camera poses. This has demonstrated remarkable success in recovering accurate camera poses from noisy pose initialization. Nonetheless, few studies have explicitly focused on the problem of LiDAR NeRF-BA, which has the potential to outperform traditional point cloud registration methods.  \nContrary to RGB images which contain rich texture for camera pose optimization, LiDAR point clouds pose the following challenges for LiDAR NeRF-BA: (1) LiDAR point clouds inherently lack textures. A direct implementation of RGB NeRF-BA on LiDAR point clouds would result in sub-optimal pose optimization, as it overlooks the strong geometric cue, i.e., accurate ranges, from LiDAR rays. (2)  \n1The authors are with Imperial College London, London, United [Kingdom.](Kingdom. ch1223@ic.ac.uk)[ ch1223@ic.ac.uk](Kingdom. ch1223@ic.ac.uk);  \n[k.zhang23@alumni.imperial.ac.uk](k.zhang23@alumni.imperial.ac.uk) ; [sen.wang@imperial.ac.uk](sen.wang@imperial.ac.uk)  \nFig. 1: 3D mapping results with HBA [5] (left) and the proposed NeLD-BA method (right) on the quad_hard sequence from the Newer College dataset [6] . NeLD-BA produces a higher quality map.  \nLiDAR point clouds are spatially sparse. Directly using the volume sampling strategy in RGB NeRFs would result in inefficient volume sampling and hinder pose optimization, which will be discussed in Section III-B.  \nTo address these challenges, we show that LiDAR NeRFBA performance is tightly coupled with volume-sampling, and propose a novel volume-sampling strategy tailored to the LiDAR ray model. Given a noisy initialization of LiDAR poses over a sequence, our algorithm, termed NeLDBA, jointly reconstructs the scene and refines the poses, outperforming existing methods in both novel view synthesis and odometry (Figure 1) .  \nOur main contributions in this work include:  \n1) We provide new insights that pose accuracy in LiDAR NeRF-BA is explicitly reinforced by volume-sampling;  \n2) We","cbCaigL39iLlaYjz","https://ap.wps.com/l/cbCaigL39iLlaYjz","pdf",19203804,1,"English","en",105,"# Introduction\n## Contributions\n# Related Work\n## Point Cloud Registration\n## LiDAR NeRF and Neural SDF","[{\"question\":\"Why can’t RGB NeRF Bundle Adjustment be directly applied to LiDAR NeRF-BA?\",\"answer\":\"LiDAR point clouds lack texture, so pose optimization would ignore the strong geometric cue of accurate LiDAR ranges. Additionally, LiDAR is spatially sparse, making standard volume sampling inefficient and detrimental to pose optimization.\"},{\"question\":\"What key factor does the paper claim is tightly coupled with LiDAR NeRF-BA performance?\",\"answer\":\"The paper argues that LiDAR NeRF-BA performance is tightly coupled with the density of volume sampling, which influences how well poses and the scene representation are jointly optimized.\"},{\"question\":\"How does NeLD-BA improve joint optimization of LiDAR maps and poses?\",\"answer\":\"NeLD-BA introduces an LiDAR-ray tailored volume sampling strategy and uses efficient volume sampling for joint optimization of the LiDAR map and poses, outperforming existing methods in both novel view synthesis and odometry.\"}]",1784197806,20,{"code":4,"msg":29,"data":30},"ok",{"site_id":23,"language":22,"slug":31,"title":13,"keywords":32,"description":14,"schema_data":33,"social_meta":84,"head_meta":86,"extra_data":88,"updated_unix":26},"neural-lidar-bundle-adjustment","",{"@graph":34,"@context":83},[35,52,66],{"@type":36,"itemListElement":37},"BreadcrumbList",[38,42,46,49],{"item":39,"name":40,"@type":41,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":43,"name":44,"@type":41,"position":45},"https://docshare.wps.com/document/","Document",2,{"item":47,"name":12,"@type":41,"position":48},"https://docshare.wps.com/document/research-report/",3,{"item":50,"name":13,"@type":41,"position":51},"https://docshare.wps.com/document/neural-lidar-bundle-adjustment/84714/",4,{"url":50,"name":13,"@type":53,"author":54,"headline":13,"publisher":56,"fileFormat":59,"inLanguage":22,"description":14,"dateModified":60,"datePublished":60,"encodingFormat":59,"isAccessibleForFree":61,"interactionStatistic":62},"DigitalDocument",{"name":9,"@type":55},"Person",{"url":39,"name":57,"@type":58},"DocShare","Organization","application/pdf","2026-07-16",true,{"@type":63,"interactionType":64,"userInteractionCount":4},"InteractionCounter",{"@type":65},"ViewAction",{"@type":67,"mainEntity":68},"FAQPage",[69,75,79],{"name":70,"@type":71,"acceptedAnswer":72},"Why can’t RGB NeRF Bundle Adjustment be directly applied to LiDAR NeRF-BA?","Question",{"text":73,"@type":74},"LiDAR point clouds lack texture, so pose optimization would ignore the strong geometric cue of accurate LiDAR ranges. Additionally, LiDAR is spatially sparse, making standard volume sampling inefficient and detrimental to pose optimization.","Answer",{"name":76,"@type":71,"acceptedAnswer":77},"What key factor does the paper claim is tightly coupled with LiDAR NeRF-BA performance?",{"text":78,"@type":74},"The paper argues that LiDAR NeRF-BA performance is tightly coupled with the density of volume sampling, which influences how well poses and the scene representation are jointly optimized.",{"name":80,"@type":71,"acceptedAnswer":81},"How does NeLD-BA improve joint optimization of LiDAR maps and poses?",{"text":82,"@type":74},"NeLD-BA introduces an LiDAR-ray tailored volume sampling strategy and uses efficient volume sampling for joint optimization of the LiDAR map and poses, outperforming existing methods in both novel view synthesis and odometry.","https://schema.org",{"og:url":50,"og:type":85,"og:title":13,"og:site_name":57,"og:description":14},"article",{"robots":87,"canonical":50},"index,follow",{"doc_id":7,"site_id":23},{"code":4,"msg":5,"data":90},[91,95,99,103,108,113,118,121,125,128,132],{"id":20,"doc_module":4,"doc_module_name":44,"category_name":92,"show_sort_weight":93,"slug":94},"Story & Novel",90,"story-novel",{"id":45,"doc_module":4,"doc_module_name":44,"category_name":96,"show_sort_weight":97,"slug":98},"Literature",80,"literature",{"id":51,"doc_module":4,"doc_module_name":44,"category_name":100,"show_sort_weight":101,"slug":102},"Exam",70,"exam",{"id":104,"doc_module":4,"doc_module_name":44,"category_name":105,"show_sort_weight":106,"slug":107},5,"Comic",60,"comic",{"id":109,"doc_module":4,"doc_module_name":44,"category_name":110,"show_sort_weight":111,"slug":112},6,"Technology",50,"technology",{"id":114,"doc_module":4,"doc_module_name":44,"category_name":115,"show_sort_weight":116,"slug":117},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":44,"category_name":12,"show_sort_weight":119,"slug":120},30,"research-report",{"id":122,"doc_module":4,"doc_module_name":44,"category_name":123,"show_sort_weight":27,"slug":124},9,"Religion & Spirituality","religion-spirituality",{"id":27,"doc_module":4,"doc_module_name":44,"category_name":126,"show_sort_weight":27,"slug":127},"World Cup","world-cup",{"id":129,"doc_module":4,"doc_module_name":44,"category_name":130,"show_sort_weight":129,"slug":131},10,"Lifestyle","lifestyle",{"id":133,"doc_module":4,"doc_module_name":44,"category_name":134,"show_sort_weight":104,"slug":135},19,"General","general"]