[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-119008-en":3,"doc-seo-119008-105":30,"detail-sidebar-cat-0-en-105":92},{"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":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},119008,549758252649,"Ivy","https://ap-avatar.wpscdn.com/avatar/8000253669c5317157?_k=1778319167496531819",8,"Research & Report","Touching a NeRF - Leveraging Neural Radiance Fields for Tactile Sensory Data Generation","Tactile perception is essential for robotic manipulation, yet tactile data collection is slow and difficult compared with vision, limiting tactile learning in robotics. This work presents a generative framework that creates realistic tactile sensory images for downstream tasks. It first trains Neural Radiance Fields (NeRF) from readily captured camera images of objects of interest, then renders RGB-D inputs for a conditional GAN that generates tactile images from target orientations. Experiments use SSIM and MSE plus tactile classification in simulation and real-world settings, showing about a 10% accuracy gain via dataset augmentation. The method also transfers across tactile sensors with minimal fine-tuning data.","Touching a NeRF: Leveraging Neural Radiance Fields for Tactile Sensory Data Generation  \nShaohong Zhong, Alessandro Albini, Oiwi Parker Jones, Perla Maiolino, Ingmar Posner  \nOxford Robotics Institute  \nUniversity of Oxford, United Kingdom  \n{shaohong,alessandro,oiwi,perla,[ingmar](ingmar}@robots.ox.ac.uk)[}](ingmar}@robots.ox.ac.uk)[@robots.ox.ac.uk](ingmar}@robots.ox.ac.uk)  \nAbstract: Tactile perception is key for robotics applications such as manipulation.  \nHowever, tactile data collection is time-consuming, especially when compared to vision. This limits the use of the tactile modality in machine learning solutions in robotics. In this paper, we propose a generative model to simulate realistic tactile sensory data for use in downstream tasks. Starting with easily-obtained camera images, we train Neural Radiance Fields (NeRF) for objects of interest.  \nWe then use NeRF-rendered RGB-D images as inputs to a conditional Generative Adversarial Network model (cGAN) to generate tactile images from desired orientations. We evaluate the generated data quantitatively using the Structural Similarity Index and Mean Squared Error metrics, and also using a tactile classification task both in simulation and in the real world. Results show that by augmenting a manually collected dataset, the generated data is able to increase classification accuracy by around 10% . In addition, we demonstrate that our model is able to transfer from one tactile sensor to another with a small fine-tuning dataset.  \nKeywords: Camera-based tactile sensing, cross-modal tactile data generation  \n1 Introduction  \nHumans rely heavily on tactile sensing for tasks such as identifying and grasping objects (e.g. picking keys from a pocket) [1, 2] . In this context, tactile sensing is fundamental to retrieve contact information or properties of the object such as roughness or stiffness, and is also able to complement vision in occluded scenarios [1] . Tactile sensing is also critical for robotics applications such as manipulation and control [3] and object or texture recognition [4] . These tasks, especially those related to tactile-based object recognition, are usually tackled with machine learning methods that typically require large amounts of data for training [1, 5, 6] . However, collecting tactile data is challenging as the robot needs to physically interact with the environment and the object. While cameras can capture the global shape of an object, tactile sensors can only capture local features, and a long and time-consuming exploration procedure is usually required to capture the whole shape [7] . Beyond the problem relating to tactile exploration, tactile sensing is also still lacking standards at the hardware level [8] . For the same physical stimulus, the output of different tactile systems can differ significantly, thus limiting the validity of the collected data to a specific sensing technology.  \nGiven the difficulties of tactile data collection, the problem of generating synthetic tactile sensor responses from data acquired using different modalities (which are easier to collect) becomes relevant. In particular, recent works show that vision data (RGB-D images) contain rich sensory information that can be used to generate tactile data [9, 10, 11] . Given a camera or depth image of the object surface as input, these approaches can generate the corresponding tactile sensor output. One limitation of these vision-based generative approaches is that they require the collection of visual samples at given positions and orientations to generate the corresponding synthetic tactile data [9] . However, with the development of neural volume-rendering techniques such as NeRF [12], it is now possible to synthesise high-quality RGB-D images for novel view orientations of a scene, given only sample 2D images and their associated camera poses [12] . In this way, NeRF provides additional information on the structure of the scene that we leverage for generating tactile imag","cbCaildKKaElBbPY","https://ap.wps.com/l/cbCaildKKaElBbPY","pdf",2988888,1,11,"English","en",105,"# Abstract\n# Introduction\n# Literature Review\n## Vision-based tactile generation\n## NeRF-enabled novel view synthesis\n# Method Overview","[{\"question\":\"Why is generating tactile data important for robotics?\",\"answer\":\"Tactile sensing underpins object identification and manipulation, but training tactile machine learning models requires large datasets that are costly to collect through physical interaction.\"},{\"question\":\"How does the method generate tactile images from camera data?\",\"answer\":\"The approach trains NeRF using camera images to synthesize RGB-D views, then feeds NeRF-rendered RGB-D into a conditional GAN to produce tactile images for desired orientations.\"},{\"question\":\"How is the generated data evaluated and what are the results?\",\"answer\":\"The study evaluates outputs using SSIM and MSE and also tests tactile classification in simulation and real-world settings. Augmenting a manually collected dataset improves classification accuracy by about 10%.\"}]","Touching a NeRF - Leveraging Neural Radiance Fields for Tactile Sensory Data Generation | PDF",1785721818,28,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":87,"head_meta":89,"extra_data":91,"updated_unix":28},"touching-a-nerf-leveraging-neural-radiance-fields-for-tactile-sensory-data-generation","",{"@graph":36,"@context":86},[37,54,69],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/touching-a-nerf-leveraging-neural-radiance-fields-for-tactile-sensory-data-generation/119008/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-04","2026-08-03",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"Why is generating tactile data important for robotics?","Question",{"text":76,"@type":77},"Tactile sensing underpins object identification and manipulation, but training tactile machine learning models requires large datasets that are costly to collect through physical interaction.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How does the method generate tactile images from camera data?",{"text":81,"@type":77},"The approach trains NeRF using camera images to synthesize RGB-D views, then feeds NeRF-rendered RGB-D into a conditional GAN to produce tactile images for desired orientations.",{"name":83,"@type":74,"acceptedAnswer":84},"How is the generated data evaluated and what are the results?",{"text":85,"@type":77},"The study evaluates outputs using SSIM and MSE and also tests tactile classification in simulation and real-world settings. Augmenting a manually collected dataset improves classification accuracy by about 10%.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,136],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":46,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":46,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":46,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":46,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":46,"category_name":138,"show_sort_weight":107,"slug":139},19,"General","general"]