{"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/auto-conditioned-recurrent-networks-for","title":"Auto-Conditioned Recurrent Networks for Extended Complex Human Motion Synthesis","arxiv_id":"1707.05363","date":"2017-07-17","proceeding":"ICLR 2018 1","authors":["Zimo Li","Yi Zhou","Shuangjiu Xiao","Chong He","Zeng Huang","Hao Li"],"abstract":"We present a real-time method for synthesizing highly complex human motions\nusing a novel training regime we call the auto-conditioned Recurrent Neural\nNetwork (acRNN). Recently, researchers have attempted to synthesize new motion\nby using autoregressive techniques, but existing methods tend to freeze or\ndiverge after a couple of seconds due to an accumulation of errors that are fed\nback into the network. Furthermore, such methods have only been shown to be\nreliable for relatively simple human motions, such as walking or running. In\ncontrast, our approach can synthesize arbitrary motions with highly complex\nstyles, including dances or martial arts in addition to locomotion. The acRNN\nis able to accomplish this by explicitly accommodating for autoregressive noise\naccumulation during training. Our work is the first to our knowledge that\ndemonstrates the ability to generate over 18,000 continuous frames (300\nseconds) of new complex human motion w.r.t. different styles.","url_abs":"http://arxiv.org/abs/1707.05363v5","url_pdf":"http://arxiv.org/pdf/1707.05363v5.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":"auto-conditioned-recurrent-networks-for","repo_url":"https://github.com/papagina/auto_conditioned_rnn_motion","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"motion-synthesis","task_name":"Motion Synthesis"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1707.05363","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}