{"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/deeply-informed-neural-sampling-for-robot","title":"Deeply Informed Neural Sampling for Robot Motion Planning","arxiv_id":"1809.10252","date":"2018-09-26","proceeding":null,"authors":["Ahmed H. Qureshi","Michael C. Yip"],"abstract":"Sampling-based Motion Planners (SMPs) have become increasingly popular as\nthey provide collision-free path solutions regardless of obstacle geometry in a\ngiven environment. However, their computational complexity increases\nsignificantly with the dimensionality of the motion planning problem. Adaptive\nsampling is one of the ways to speed up SMPs by sampling a particular region of\na configuration space that is more likely to contain an optimal path solution.\nAlthough there are a wide variety of algorithms for adaptive sampling, they\nrely on hand-crafted heuristics; furthermore, their performance decreases\nsignificantly in high-dimensional spaces. In this paper, we present a neural\nnetwork-based adaptive sampler for motion planning called Deep Sampling-based\nMotion Planner (DeepSMP). DeepSMP generates samples for SMPs and enhances their\noverall speed significantly while exhibiting efficient scalability to\nhigher-dimensional problems. DeepSMP's neural architecture comprises of a\nContractive AutoEncoder which encodes given workspaces directly from a raw\npoint cloud data, and a Dropout-based stochastic deep feedforward neural\nnetwork which takes the workspace encoding, start and goal configuration, and\niteratively generates feasible samples for SMPs to compute end-to-end\ncollision-free optimal paths. DeepSMP is not only consistently computationally\nefficient in all tested environments but has also shown remarkable\ngeneralization to completely unseen environments. We evaluate DeepSMP on\nmultiple planning problems including planning of a point-mass robot,\nrigid-body, 6-link robotic manipulator in various 2D and 3D environments. The\nresults show that on average our method is at least 7 times faster in\npoint-mass and rigid-body case and about 28 times faster in 6-link robot case\nthan the existing state-of-the-art.","url_abs":"http://arxiv.org/abs/1809.10252v1","url_pdf":"http://arxiv.org/pdf/1809.10252v1.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":"deeply-informed-neural-sampling-for-robot","repo_url":"https://github.com/ahq1993/MPNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"motion-planning","task_name":"Motion Planning"}],"methods":[{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1809.10252","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}