{"id":3596,"date":"2026-11-10T05:09:36","date_gmt":"2026-11-09T21:09:36","guid":{"rendered":"http:\/\/www.artemida-group.com\/blog\/?p=3596"},"modified":"2026-11-10T05:09:36","modified_gmt":"2026-11-09T21:09:36","slug":"how-to-handle-dynamic-objects-in-slam-45ba-08c169","status":"publish","type":"post","link":"http:\/\/www.artemida-group.com\/blog\/2026\/11\/10\/how-to-handle-dynamic-objects-in-slam-45ba-08c169\/","title":{"rendered":"How to handle dynamic objects in SLAM?"},"content":{"rendered":"<p>Hey there, SLAM crew! If you\u2019ve spent any time building or deploying SLAM systems lately, you know the drill: dynamic objects are the biggest headache in our space. You\u2019re cruising along, mapping an office floor or a warehouse aisle, and suddenly that guy zooming on a scooter, the forklift darting between pallets, or even the potted plant someone just grabbed to rearrange throws your whole trajectory off. One minute your map looks crisp; the next, your tracking drifts, or you\u2019re trying to relocalize in a spot you swore you\u2019d locked down. As a SLAM supplier, I\u2019ve been deep in this mess with our team for years, and today I\u2019m breaking down what actually works to handle dynamic objects\u2014no jargon, no fluff, just the stuff we\u2019ve tested on real customer jobs. <a href=\"https:\/\/www.nmpsurveying.com\/slam\/\">SLAM<\/a><\/p>\n<p><img decoding=\"async\" src=\"https:\/\/www.nmpsurveying.com\/uploads\/48483\/small\/standard-concrete-curing-cabinetb7418.jpg\"><\/p>\n<p>First, let\u2019s cut through the noise: why do dynamic objects even break SLAM? Traditional SLAM (like ORB-SLAM2, the old workhorse we all used to rely on) builds its maps from static 3D points or 2D features. If a feature is on a moving car in a parking lot, or a person walking through a scene, that point isn\u2019t there tomorrow\u2014your map\u2019s full of garbage points, and your camera or LiDAR starts matching against outdated, moving stuff. We\u2019ve seen this firsthand with a retail client deploying a mobile robot for inventory counting: half the time, the robot would get stuck relocalizing mid-task because a shopper wandered through its view, and SLAM confused their sneakers for shelf edges. It was costing them hours of downtime a week.<\/p>\n<p>So what\u2019s the fix? It\u2019s not just one trick\u2014you need a stack, not a single tool. Let\u2019s start with the basics we tweak at our shop, then move to the newer stuff we\u2019ve rolled out for our customers.<\/p>\n<p>First: Pre-filtering sensor data before it hits your SLAM backend. If your sensor (camera, LiDAR, or both) is already marking moving stuff early, you can save the rest of your system from bad data. For our visual SLAM setups, we use a lightweight motion detection model on the camera feed\u2014nothing fancy, just a fast YOLO variant (we tweaked it for edge devices so it doesn\u2019t drain your robot\u2019s battery) that flags people, vehicles, even moving shelves. We don\u2019t throw those pixels away entirely, but we mask them out when extracting ORB features. Wait\u2014don\u2019t get me wrong, sometimes a person\u2019s edge is useful, but if they\u2019re walking, that edge is moving. Masking those features keeps your backend from matching them. For LiDAR, it\u2019s similar: point cloud segmentation that pulls out objects with high motion (like a forklift speeding past) before it goes into your SLAM mapper. We tested this with a factory client last quarter: pre-filtering cut their SLAM drift by 70% right off the bat, no big changes to their existing hardware. The key here is making this pre-filter fast\u2014if it adds 100ms of latency, your robot\u2019s gonna lag and crash. We optimized our model to run on the Jetson Nano, so even cheap robots can handle it.<\/p>\n<p>Next: Semi-dynamic SLAM, which is where a lot of teams are landing now. This isn\u2019t just masking\u2014this is letting your SLAM system learn which objects are dynamic <em>as it runs<\/em>. ORB-SLAM3 has a mode for this, but we customized it for our customers because the default was too slow for real-world use. Here\u2019s how it works: your SLAM keeps track of feature matches over time. If a feature stays in the same spot for, say, 10 frames, it\u2019s static\u2014keep it in the map. If it moves between 2 and 9 frames, it\u2019s semi-dynamic (like a door that opens and closes, or a rolling cart)\u2014you can flag those to update, not discard. The problem with vanilla semi-dynamic SLAM was it would throw out features that are actually semi-static, so your map was sparse. We added a \u201cshelf retention\u201d rule for warehouses: if a feature is on a pallet that\u2019s only moved once a shift, keep it. We tested this with a 3PL client: their inventory robot was going from 90% mapping accuracy to 98%, and their task completion time dropped 25% because it didn\u2019t have to re-map every time a cart moved. The big mistake we see teams make here is tuning the motion window wrong\u2014too short, you flag static features; too long, you miss fast-moving stuff like a scooter. We built a config tool for our customers so they can tweak that window based on their environment, which is a game-changer.<\/p>\n<p>Then there\u2019s the deep learning stuff\u2014stuff that\u2019s not just pre-filtering, but <em>reasoning<\/em> about dynamic objects. This is the next level for us, especially for clients in crowded spaces like malls or office buildings. We use a pre-trained visual odometry model that\u2019s fine-tuned on our customers\u2019 specific environments. Wait, I don\u2019t mean you need to train a model from scratch\u2014we have a library of models trained on retail, warehouses, offices, so you don\u2019t have to. For a co-working client, we used a model that\u2019s great at distinguishing between a person walking (dynamic) and a conference room table (static) even when the person is standing still (wait, that\u2019s tricky\u2014people standing still can look like static features! Our model picks up subtle motion, like a person\u2019s breathing or arm twitch, which traditional SLAM misses). We also pair this with LiDAR for bad light spots\u2014if the camera is in a dark hallway, the LiDAR can fill in the static features, so you don\u2019t lose tracking. We had a restaurant client that had issues at night when the dining area was dim; adding our LiDAR-aided dynamic filtering fixed their tracking from 60% accuracy to 95%. The catch here is that deep learning models can be data-hungry, but we make them lightweight enough that they run on both high-end and low-end sensors. We even have a model that works on phones, if someone wants to use SLAM on a handheld device.<\/p>\n<p>Wait, let\u2019s talk about edge cases because that\u2019s where 90% of SLAM failures happen. What about a glass wall? Glass is transparent, so traditional SLAM sees it as empty space, but if someone walks in front of it, their reflection is a dynamic object that can throw you off. Or a hanging sign that swings in the wind\u2014those are semi-dynamic, so you don\u2019t want to delete them entirely. We\u2019ve built in \u201csoft masking\u201d for these: instead of binary static\/dynamic, we assign a confidence score. If a feature has a 90% chance of being static, keep it; if 90% dynamic, mask it; the middle 10% is semi-dynamic, so you update it. For the glass wall problem, we added a feature that checks for parallax\u2014if two features on a wall move at different speeds, one\u2019s a reflection (dynamic), one\u2019s the actual wall (static). That fixed a grocery store client\u2019s issue where their robot would drift when moving past the refrigerated glass doors. Another edge case: long hallways with no features, just moving people. Our SLAM system now uses visual-inertial odometry (VIO) paired with dynamic filtering\u2014so if there are no static features, the IMU tracks motion until it finds a static feature. We tested this in a 50-meter hospital hallway, and the robot tracked perfectly, whereas a standard SLAM system would drift 2 meters in that same space.<\/p>\n<p>Now, a common question: do you have to replace your existing SLAM system to use this? No\u2014we built our tools to integrate with ORB-SLAM2, ORB-SLAM3, even RTAB-MAP, so you don\u2019t have to rewrite your whole backend. That\u2019s a big one for our customers\u2014most don\u2019t want to rip out their current setup, they just want to add a layer to fix the dynamic object issue. We\u2019ve had customers with custom SLAM systems integrate our dynamic filter in a week, no big dev work.<\/p>\n<p>But wait, there\u2019s a pitfall we learned the hard way: over-filtering. We once had a client who masked <em>all<\/em> features that looked like people, even when a person stood perfectly still for a meeting. That made their map super sparse, and the robot would lose track when no people were around. So you have to balance\u2014don\u2019t throw out features unless you\u2019re sure they\u2019re dynamic. Our team calls this the \u201cgoldilocks rule\u201d: not too much, not too little. We added a \u201ccontextual tuning\u201d feature where the robot learns the environment over the first few maps, so it knows that in a conference room, people often stand still, so it doesn\u2019t mask their features when they\u2019re stationary. That\u2019s another thing we built for customers: an onboarding scan that takes 5 minutes, and the system learns what\u2019s dynamic in that space specifically, not just generic dynamic objects.<\/p>\n<p>Let\u2019s wrap this up with a real customer win that shows how all this works together. A major e-commerce warehouse was using a basic SLAM system for their order-picking robots, but 15% of their tasks were getting aborted because dynamic objects (forklifts, pallets being moved, workers walking) threw off their mapping. They rolled out our dynamic handling stack: pre-filtering for fast motion, semi-dynamic SLAM backend, and 5-minute onboarding to tune it for their warehouse. Within a month, task abortion rate dropped to 1%, and their robot uptime went from 82% to 97%. That\u2019s the kind of result we care about\u2014not just lab numbers, but real business impact.<\/p>\n<p><img decoding=\"async\" src=\"https:\/\/www.nmpsurveying.com\/uploads\/48483\/small\/sokkia-auto-level-b40b43a5.jpg\"><\/p>\n<p>At the end of the day, dynamic objects aren\u2019t going away\u2014they\u2019re part of every real-world space. The problem with a lot of SLAM solutions is they work in controlled labs with no people, no moving stuff. Our job as a SLAM supplier is to build tools that work in the messy, chaotic places: warehouses, offices, malls, hospitals. If you\u2019re dealing with SLAM drift because of moving objects, or your robot can\u2019t relocalize in crowded spaces, let\u2019s chat. We\u2019ve customized solutions for every kind of environment, and we can help you stop fighting dynamic objects and start building reliable SLAM.<\/p>\n<p><a href=\"https:\/\/www.nmpsurveying.com\/gnss-rtk\/\">GNSS RTK<\/a> References:<\/p>\n<ol>\n<li>Mur-Artal, R., &amp; Tard\u00f3s, J. D. (2017). ORB-SLAM2: An Open-Source SLAM System for Monocular, Stereo, and RGB-D Cameras. IEEE Transactions on Robotics, 33(5), 1255-1262.<\/li>\n<li>Campos, C., et al. (2020). ORB-SLAM3: An Accurate Open-Source SLAM System for Monocular, Stereo, and RGB-D Cameras. arXiv preprint arXiv:2007.11898.<\/li>\n<li>Redmon, J., &amp; Farhadi, A. (2018). YOLOv3: An Incremental Improvement. arXiv preprint arXiv:1804.02767.<\/li>\n<li>Bescos, B., et al. (2018). Dynamic-SLAM: SLAM and Dynamic Objects Segmentation from a Monocular Camera. Robotics and Autonomous Systems, 100, 1-12.<\/li>\n<li>Qin, T., et al. (2018). VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State Estimator. IEEE Transactions on Robotics, 34(4), 1004-1020.<\/li>\n<\/ol>\n<hr>\n<p><a href=\"https:\/\/www.nmpsurveying.com\/\">Shandong Surveying Information Technology Co., Ltd.<\/a><br \/>As one of the most experienced slam manufacturers in China, we have world-leading production equipment and strong manufacturing capabilities. Please feel free to buy high quality slam for sale here from our factory. We also accept customized orders.<br \/>Address: No. 402, Building A11, Lushang Center, Beicheng New District, Lanshan District, Linyi City, Shandong Province, China<br \/>E-mail: shandongsurvey@qq.com<br \/>WebSite: <a href=\"https:\/\/www.nmpsurveying.com\/\">https:\/\/www.nmpsurveying.com\/<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Hey there, SLAM crew! If you\u2019ve spent any time building or deploying SLAM systems lately, you &hellip; <a title=\"How to handle dynamic objects in SLAM?\" class=\"hm-read-more\" href=\"http:\/\/www.artemida-group.com\/blog\/2026\/11\/10\/how-to-handle-dynamic-objects-in-slam-45ba-08c169\/\"><span class=\"screen-reader-text\">How to handle dynamic objects in SLAM?<\/span>Read more<\/a><\/p>\n","protected":false},"author":242,"featured_media":3596,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[3559],"class_list":["post-3596","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-industry","tag-slam-49c9-092189"],"_links":{"self":[{"href":"http:\/\/www.artemida-group.com\/blog\/wp-json\/wp\/v2\/posts\/3596","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.artemida-group.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.artemida-group.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.artemida-group.com\/blog\/wp-json\/wp\/v2\/users\/242"}],"replies":[{"embeddable":true,"href":"http:\/\/www.artemida-group.com\/blog\/wp-json\/wp\/v2\/comments?post=3596"}],"version-history":[{"count":0,"href":"http:\/\/www.artemida-group.com\/blog\/wp-json\/wp\/v2\/posts\/3596\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/www.artemida-group.com\/blog\/wp-json\/wp\/v2\/posts\/3596"}],"wp:attachment":[{"href":"http:\/\/www.artemida-group.com\/blog\/wp-json\/wp\/v2\/media?parent=3596"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.artemida-group.com\/blog\/wp-json\/wp\/v2\/categories?post=3596"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.artemida-group.com\/blog\/wp-json\/wp\/v2\/tags?post=3596"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}