{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/visual-global-localization-with-a-hybrid-wnn","title":"Visual Global Localization with a Hybrid WNN-CNN Approach","arxiv_id":"1805.03183","date":"2018-05-08","proceeding":null,"authors":["Avelino Forechi","Thiago Oliveira-Santos","Claudine Badue","Alberto F. de Souza"],"abstract":"Currently, self-driving cars rely greatly on the Global Positioning System\n(GPS) infrastructure, albeit there is an increasing demand for alternative\nmethods for GPS-denied environments. One of them is known as place recognition,\nwhich associates images of places with their corresponding positions. We\npreviously proposed systems based on Weightless Neural Networks (WNN) to\naddress this problem as a classification task. This encompasses solely one part\nof the global localization, which is not precise enough for driverless cars.\nInstead of just recognizing past places and outputting their poses, it is\ndesired that a global localization system estimates the pose of current place\nimages. In this paper, we propose to tackle this problem as follows. Firstly,\ngiven a live image, the place recognition system returns the most similar image\nand its pose. Then, given live and recollected images, a visual localization\nsystem outputs the relative camera pose represented by those images. To\nestimate the relative camera pose between the recollected and the current\nimages, a Convolutional Neural Network (CNN) is trained with the two images as\ninput and a relative pose vector as output. Together, these systems solve the\nglobal localization problem using the topological and metric information to\napproximate the current vehicle pose. The full approach is compared to a Real-\nTime Kinematic GPS system and a Simultaneous Localization and Mapping (SLAM)\nsystem. Experimental results show that the proposed approach correctly\nlocalizes a vehicle 90% of the time with a mean error of 1.20m compared to\n1.12m of the SLAM system and 0.37m of the GPS, 89% of the time.","url_abs":"http://arxiv.org/abs/1805.03183v2","url_pdf":"http://arxiv.org/pdf/1805.03183v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"visual-global-localization-with-a-hybrid-wnn","repo_url":"https://github.com/LCAD-UFES/WNN-CNN-GL","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"self-driving-cars","task_name":"Self-Driving Cars"},{"task_slug":"simultaneous-localization-and-mapping","task_name":"Simultaneous Localization and Mapping"},{"task_slug":"visual-localization","task_name":"Visual Localization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}