{"id":21026,"date":"2026-08-29T15:27:27","date_gmt":"2026-08-29T15:27:27","guid":{"rendered":"https:\/\/8657085.xyz\/?p=21026"},"modified":"2026-08-29T15:27:27","modified_gmt":"2026-08-29T15:27:27","slug":"ai-learns-to-model-motion-more-intelligently-new-framework-improves-3d-reconstruction-of-complex-dynamic-scenes","status":"publish","type":"post","link":"https:\/\/8657085.xyz\/?p=21026","title":{"rendered":"AI learns to model motion more intelligently: New framework improves 3D reconstruction of complex dynamic scenes"},"content":{"rendered":"<p> <div style=\"display: grid; grid-template-columns: 300px 160px; gap: 2px; width: 460px; background: #eee; padding: 2px;\">\r\n\r\n  <!-- \u6574\u884c\u5bbd\u5e7f\u544a -->\r\n  <div style=\"grid-column: 1\/-1; width: 460px; height: 250px; background: #ccc; display: grid; place-items: center;\">\r\n  <script async type=\"application\/javascript\" src=\"https:\/\/a.magsrv.com\/ad-provider.js\"><\/script> \r\n <ins class=\"eas6a97888e2\" data-zoneid=\"5876674\"><\/ins> \r\n <script>(AdProvider = window.AdProvider || []).push({\"serve\": {}});<\/script>\r\n  <\/div>\r\n  <div style=\"grid-column: 1\/-1; width: 460px; height: 90px; background: #ccc; display: grid; place-items: center;\">\r\n  <script async type=\"application\/javascript\" src=\"https:\/\/a.magsrv.com\/ad-provider.js\"><\/script> \r\n <ins class=\"eas6a97888e2\" data-zoneid=\"5876676\"><\/ins> \r\n <script>(AdProvider = window.AdProvider || []).push({\"serve\": {}});<\/script>\r\n  <\/div>\r\n\r\n  <!-- \u5de6\u4fa7\u7ad6\u6392 -->\r\n  <div style=\"height: 250px; background: #ccc; display: grid; place-items: center;\">\r\n  <script async type=\"application\/javascript\" src=\"https:\/\/a.magsrv.com\/ad-provider.js\"><\/script> \r\n <ins class=\"eas6a97888e2\" data-zoneid=\"5876672\"><\/ins> \r\n <script>(AdProvider = window.AdProvider || []).push({\"serve\": {}});<\/script>\r\n  <\/div>\r\n  <div style=\"height: 500px; background: #ccc; display: grid; place-items: center;\">\r\n  <script async type=\"application\/javascript\" src=\"https:\/\/a.magsrv.com\/ad-provider.js\"><\/script> \r\n <ins class=\"eas6a97888e2\" data-zoneid=\"5876680\"><\/ins> \r\n <script>(AdProvider = window.AdProvider || []).push({\"serve\": {}});<\/script>\r\n  <\/div>\r\n\r\n  <!-- \u53f3\u4fa7\u6469\u5929\u697c\uff08\u548c\u5de6\u4fa7\u5b8c\u5168\u5bf9\u9f50\uff09 -->\r\n  <div style=\"grid-row: 3\/5; height: 750px; background: #ccc; display: grid; place-items: center;\">\r\n  <script async type=\"application\/javascript\" src=\"https:\/\/a.magsrv.com\/ad-provider.js\"><\/script> \r\n <ins class=\"eas6a97888e2\" data-zoneid=\"5876678\"><\/ins> \r\n <script>(AdProvider = window.AdProvider || []).push({\"serve\": {}});<\/script>\r\n  <\/div>\r\n  \r\n  <script async type=\"application\/javascript\" src=\"https:\/\/a.magsrv.com\/ad-provider.js\"><\/script> \r\n <ins class=\"eas6a97888e6\" data-zoneid=\"5876682\"><\/ins> \r\n <script>(AdProvider = window.AdProvider || []).push({\"serve\": {}});<\/script>\r\n<\/div><br \/>\n<\/p>\n<div style=\"padding-right:0;padding-left:0\">\n<p class=\"wp-block-paragraph\">One of the challenges facing artificial intelligence is teaching machines to understand a world that is constantly moving. From self-driving cars navigating busy streets to digital twins replicating factories in real time, AI systems need accurate ways to reconstruct and interpret dynamic three-dimensional environments. A new study led by researchers at South Korea\u2019s Pusan National University suggests that the answer may lie not in finding a single perfect motion model, but in combining several specialized models that each excel at understanding different types of movement. The research introduces two new frameworks, called <strong>MoE-GS<\/strong> and <strong>MoDE<\/strong>, that significantly improve the reconstruction of dynamic 3D scenes by leveraging a \u201cmixture of experts\u201d approach. The findings were published in <em>IEEE Transactions on Pattern Analysis and Machine Intelligence<\/em> (TPAMI). [pusan.ac.kr], [prnewswire.com]<\/p>\n<p class=\"wp-block-paragraph\">The development could have important implications for robotics, autonomous vehicles, spatial computing, virtual and augmented reality, industrial digital twins, and emerging Physical AI systems that interact directly with the real world. <\/p>\n<h2 id=\"h-why-reconstructing-motion-is-difficult\" class=\"wp-block-heading\">Why reconstructing motion is difficult<\/h2>\n<p class=\"wp-block-paragraph\">Modern computer vision systems have become remarkably capable at reconstructing static environments. Technologies such as 3D Gaussian Splatting allow computers to represent objects and spaces with impressive realism and efficiency. However, when scenes become dynamic, matters become considerably more complicated.<\/p>\n<p class=\"wp-block-paragraph\">Real-world environments contain many forms of motion occurring simultaneously. A pedestrian may be walking while a vehicle is accelerating, leaves are blowing in the wind, and shadows are shifting across the ground. Different mathematical approaches often handle certain types of motion well while struggling with others. According to the Pusan National University team, their systematic analysis revealed that no existing Dynamic Gaussian Splatting (DGS) approach consistently outperforms alternatives across all dynamic scenarios. Each method possesses strengths and weaknesses depending on the specific characteristics of the movement being modelled. <\/p>\n<p class=\"wp-block-paragraph\">Lead researcher Professor Kyeongbo Kong explained: \u201cOur findings suggest that combining multiple specialized motion representations can be an effective way to handle heterogeneous dynamics that are difficult for a single representation to model consistently.\u201d <\/p>\n<h2 id=\"h-moving-beyond-one-size-fits-all-ai\" class=\"wp-block-heading\">Moving beyond one-size-fits-all AI<\/h2>\n<p class=\"wp-block-paragraph\">The researchers\u2019 solution draws inspiration from a concept that has become increasingly influential in artificial intelligence: the <strong>Mixture of Experts (MoE)<\/strong> architecture. Rather than relying on a single model to perform every task, Mixture of Experts systems employ several specialized components, each optimized for particular challenges. The system then determines which expert is best suited to handle a given situation. This principle has already been applied successfully in large language models. The Pusan team sought to adapt the concept to dynamic 3D scene reconstruction. <\/p>\n<p class=\"wp-block-paragraph\">The result was the creation of two complementary frameworks. The first, <strong>MoE-GS (Mixture of Experts for Dynamic Gaussian Splatting)<\/strong>, independently trains multiple dynamic Gaussian models and then intelligently blends their outputs using what the researchers call a \u201cVolume-aware Pixel Router.\u201d This adaptive routing system determines which model\u2019s predictions are most appropriate for different parts of a scene and different moments in time. The second framework, <strong>MoDE (Mixture of Deformation Experts)<\/strong>, takes a different approach by training multiple motion experts together within a shared Gaussian representation. Instead of combining fully independent models after training, the system learns to capture multiple motion patterns simultaneously within a unified structure. <\/p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" width=\"1200\" height=\"708\" src=\"https:\/\/www.digitaljournal.com\/wp-content\/uploads\/2026\/08\/purcs_213_7_purcs_213_infographic_final-1200x708.jpg\" alt=\"\" class=\"wp-image-3861397\" srcset=\"https:\/\/www.digitaljournal.com\/wp-content\/uploads\/2026\/08\/purcs_213_7_purcs_213_infographic_final-1200x708.jpg 1200w, https:\/\/www.digitaljournal.com\/wp-content\/uploads\/2026\/08\/purcs_213_7_purcs_213_infographic_final-800x472.jpg 800w, https:\/\/www.digitaljournal.com\/wp-content\/uploads\/2026\/08\/purcs_213_7_purcs_213_infographic_final-768x453.jpg 768w, https:\/\/www.digitaljournal.com\/wp-content\/uploads\/2026\/08\/purcs_213_7_purcs_213_infographic_final-1536x906.jpg 1536w, https:\/\/www.digitaljournal.com\/wp-content\/uploads\/2026\/08\/purcs_213_7_purcs_213_infographic_final-2048x1208.jpg 2048w\" sizes=\"(max-width: 1200px) 100vw, 1200px\"\/><figcaption class=\"wp-element-caption\">Researchers developed two Mixture-of-Experts strategies that combine specialized dynamic  representations to reconstruct complex dynamic scenes more accurately than conventional single-model<br \/>approaches. Image: Professor Kyeongbo Kong \/ Pusan National University, with permisson.<\/figcaption><\/figure>\n<p class=\"wp-block-paragraph\">The key insight behind both frameworks is that different kinds of movement require different forms of expertise. Some motion models excel at representing rigid motion, such as moving vehicles or machinery. Others perform better when modelling flexible or deformable structures such as people, animals, cloth, or vegetation.<\/p>\n<p class=\"wp-block-paragraph\">Traditional approaches attempt to force a single representation to accommodate all possibilities. The new frameworks instead embrace specialization and combine multiple experts where appropriate. Experimental evaluations demonstrated that these multi-expert approaches can reconstruct complex dynamic scenes more accurately than conventional single-model systems. The improvements were achieved while exploring various trade-offs between performance, computational efficiency, and model flexibility. The researchers believe this represents a more scalable pathway for future dynamic scene understanding systems.<\/p>\n<p class=\"wp-block-paragraph\">Perhaps the most immediate application lies in robotics and autonomous vehicles. Self-driving cars must continuously construct an accurate representation of their surroundings while interpreting multiple moving objects in real time. Errors in scene reconstruction can lead to incorrect predictions and potentially dangerous decisions. By providing more reliable modelling of diverse motion patterns, multi-expert frameworks could help future autonomous systems better understand complex traffic conditions, crowded urban environments, and rapidly changing situations.  The same principle applies to mobile robots operating in warehouses, factories, hospitals, and public spaces. As AI systems become increasingly responsible for interacting with dynamic environments, robust motion understanding becomes essential.<\/p>\n<h2 id=\"h-advancing-digital-twins-and-spatial-computing\" class=\"wp-block-heading\">Advancing digital twins and spatial computing<\/h2>\n<p class=\"wp-block-paragraph\">Another major beneficiary could be the rapidly growing digital twin sector. Digital twins are virtual replicas of physical assets, facilities, and entire environments. They are increasingly used in manufacturing, urban planning, logistics, and infrastructure management. Maintaining an accurate digital twin requires continual updating of dynamic changes occurring in the real world. More accurate reconstruction technologies could significantly improve the fidelity and usefulness of such systems.  Similarly, spatial computing applications, including virtual reality (VR), augmented reality (AR), and mixed reality systems, require precise environmental modelling to maintain immersion and realism. As technology companies invest heavily in next-generation immersive experiences, advances in dynamic scene reconstruction may become increasingly important.<\/p>\n<p class=\"wp-block-paragraph\">The study also touches upon one of the fastest-growing areas in artificial intelligence: Physical AI. Unlike conventional AI systems that operate primarily in digital environments, Physical AI interacts directly with the physical world through robots, sensors, and autonomous machines. To function effectively, these systems must maintain accurate world models capable of interpreting continuously changing surroundings. The Pusan framework contributes to this broader goal by improving a machine\u2019s ability to understand movement, predict changes, and update its representation of reality.  While the technology remains largely within the research domain, its potential applications span numerous industries.<\/p>\n<p class=\"wp-block-paragraph\">For years, many AI advances have focused on building larger and more powerful individual models. The success of MoE-GS and MoDE suggests that future progress may increasingly come from combining specialized intelligence rather than pursuing a universal solution.<\/p>\n<\/div>\n<p><!-- \u603b\u5bb9\u5668\uff1a\u6700\u5927\u5bbd908px Grid\u7d27\u51d1\u5e03\u5c40 -->\r\n<div style=\"display: grid; grid-template-columns: 728px 160px; gap:2px; width:908px; background:#eee; padding:2px;\">\r\n\r\n  <!-- \u901a\u680f\u9876\u90e8\uff1a\u6700\u5927\u6a2a\u5e45 908x258 \u8de8\u6574\u884c -->\r\n  <div style=\"grid-column:1\/-1; height:258px; background:#ff6b6b; display:grid; place-items:center;\">\r\n    <!-- JuicyAds v3.0 -->\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async src=\"https:\/\/poweredby.jads.co\/js\/jads.js\"><\/script>\r\n<ins id=\"1114307\" data-width=\"908\" data-height=\"258\"><\/ins>\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async>(adsbyjuicy = window.adsbyjuicy || []).push({'adzone':1114307});<\/script>\r\n<!--JuicyAds END-->\r\n  <\/div>\r\n\r\n  <!-- \u7b2c\u4e8c\u901a\u680f\uff1a728\u00d790 \u901a\u680f -->\r\n  <div style=\"grid-column:1\/-1; height:90px; background:#4ecdc4; display:grid; place-items:center;\">\r\n    <!-- JuicyAds v3.0 -->\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async src=\"https:\/\/poweredby.jads.co\/js\/jads.js\"><\/script>\r\n<ins id=\"1114300\" data-width=\"728\" data-height=\"90\"><\/ins>\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async>(adsbyjuicy = window.adsbyjuicy || []).push({'adzone':1114300});<\/script>\r\n<!--JuicyAds END-->\r\n  <\/div>\r\n\r\n  <!-- \u5de6\u4fa7\u4e3b\u680f\uff1a\u591a\u5e7f\u544a\u5806\u53e0 -->\r\n  <div style=\"display:grid; gap:2px;\">\r\n    <div style=\"height:60px; background:#45b7d1; display:grid; place-items:center;\">\r\n\t<!-- JuicyAds v3.0 -->\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async src=\"https:\/\/poweredby.jads.co\/js\/jads.js\"><\/script>\r\n<ins id=\"1114308\" data-width=\"468\" data-height=\"60\"><\/ins>\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async>(adsbyjuicy = window.adsbyjuicy || []).push({'adzone':1114308});<\/script>\r\n<!--JuicyAds END-->\r\n\t<\/div>\r\n    <div style=\"height:250px; background:#ffe066; display:grid; place-items:center;\">\r\n\t<!-- JuicyAds v3.0 -->\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async src=\"https:\/\/poweredby.jads.co\/js\/jads.js\"><\/script>\r\n<ins id=\"1114299\" data-width=\"300\" data-height=\"250\"><\/ins>\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async>(adsbyjuicy = window.adsbyjuicy || []).push({'adzone':1114299});<\/script>\r\n<!--JuicyAds END-->\r\n\t<\/div>\r\n    <div style=\"height:250px; background:#ff9ecd; display:grid; place-items:center;\">\r\n\t<!-- JuicyAds v3.0 -->\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async src=\"https:\/\/poweredby.jads.co\/js\/jads.js\"><\/script>\r\n<ins id=\"1114305\" data-width=\"250\" data-height=\"250\"><\/ins>\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async>(adsbyjuicy = window.adsbyjuicy || []).push({'adzone':1114305});<\/script>\r\n<!--JuicyAds END-->\r\n\t<\/div>\r\n    <div style=\"height:139px; background:#c792ea; display:grid; place-items:center;\">\r\n\t<!-- JuicyAds v3.0 -->\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async src=\"https:\/\/poweredby.jads.co\/js\/jads.js\"><\/script>\r\n<ins id=\"1114302\" data-width=\"133\" data-height=\"139\"><\/ins>\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async>(adsbyjuicy = window.adsbyjuicy || []).push({'adzone':1114302});<\/script>\r\n<!--JuicyAds END-->\r\n\t<\/div>\r\n    <div style=\"height:125px; background:#91e7ac; display:grid; place-items:center;\">\r\n\t\r\n<!-- JuicyAds v3.0 -->\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async src=\"https:\/\/poweredby.jads.co\/js\/jads.js\"><\/script>\r\n<ins id=\"1114303\" data-width=\"125\" data-height=\"125\"><\/ins>\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async>(adsbyjuicy = window.adsbyjuicy || []).push({'adzone':1114303});<\/script>\r\n<!--JuicyAds END-->\r\n\t<\/div>\r\n  <\/div>\r\n\r\n  <!-- \u53f3\u4fa7\u7ad6\u680f\uff1a160\u00d7600 \u6574\u5217\u9ad8\u5e7f\u544a -->\r\n  <div style=\"grid-row:3\/8; height:600px;  display:grid; place-items:center;\">\r\n    <!-- JuicyAds v3.0 -->\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async src=\"https:\/\/poweredby.jads.co\/js\/jads.js\"><\/script>\r\n<ins id=\"1114301\" data-width=\"160\" data-height=\"600\"><\/ins>\r\n<script type=\"text\/javascript\" data-cfasync=\"false\" async>(adsbyjuicy = window.adsbyjuicy || []).push({'adzone':1114301});<\/script>\r\n<!--JuicyAds END-->\r\n  <\/div>\r\n\r\n<\/div><br \/>\n<br \/> AI learns to model motion more intelligently: New framework improves 3D reconstruction of complex dynamic scenes<br \/>\n<br \/>#learns #model #motion #intelligently #framework #improves #reconstruction #complex #dynamic #scenes<\/p>\n","protected":false},"excerpt":{"rendered":"<p>One of the challenges facing artificial intelligence is teaching machines to understand a world that&#8230;<\/p>\n","protected":false},"author":1,"featured_media":21027,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[7],"tags":[1915,15624,2985,10575,15623,4362,3333,5384,11470,7550],"class_list":["post-21026","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-stories","tag-complex","tag-dynamic","tag-framework","tag-improves","tag-intelligently","tag-learns","tag-model","tag-motion","tag-reconstruction","tag-scenes"],"featured_image_urls":{"full":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT.jpg",1151,863,false],"thumbnail":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT-150x150.jpg",150,150,true],"medium":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT-300x225.jpg",300,225,true],"medium_large":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT-768x576.jpg",640,480,true],"large":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT-1024x768.jpg",640,480,true],"1536x1536":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT.jpg",1151,863,false],"2048x2048":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT.jpg",1151,863,false],"covernews-slider-full":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT-1115x715.jpg",1115,715,true],"covernews-slider-center":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT-800x500.jpg",800,500,true],"covernews-featured":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT-1024x768.jpg",1024,768,true],"covernews-medium":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT-540x340.jpg",540,340,true],"covernews-medium-square":["https:\/\/8657085.xyz\/wp-content\/uploads\/2026\/08\/20220305_165412-EDIT-400x250.jpg",400,250,true]},"author_info":{"display_name":"admin","author_link":"https:\/\/8657085.xyz\/?author=1"},"category_info":"<a href=\"https:\/\/8657085.xyz\/?cat=7\" rel=\"category\">Stories<\/a>","tag_info":"Stories","comment_count":"0","_links":{"self":[{"href":"https:\/\/8657085.xyz\/index.php?rest_route=\/wp\/v2\/posts\/21026","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/8657085.xyz\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/8657085.xyz\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/8657085.xyz\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/8657085.xyz\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=21026"}],"version-history":[{"count":0,"href":"https:\/\/8657085.xyz\/index.php?rest_route=\/wp\/v2\/posts\/21026\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/8657085.xyz\/index.php?rest_route=\/wp\/v2\/media\/21027"}],"wp:attachment":[{"href":"https:\/\/8657085.xyz\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=21026"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/8657085.xyz\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=21026"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/8657085.xyz\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=21026"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}