{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,7]],"date-time":"2026-04-07T08:44:11Z","timestamp":1775551451705,"version":"3.50.1"},"reference-count":49,"publisher":"American Association for the Advancement of Science (AAAS)","issue":"105","content-domain":{"domain":["www.science.org"],"crossmark-restriction":true},"short-container-title":["Sci. Robot."],"published-print":{"date-parts":[[2025,8,27]]},"abstract":"<jats:p>Dynamic locomotion of legged robots is a critical yet challenging topic in expanding the operational range of mobile robots. It requires precise planning when possible footholds are sparse, robustness against uncertainties and disturbances, and generalizability across diverse terrains. Although traditional model-based controllers excel at planning on complex terrains, they struggle with real-world uncertainties. Learning-based controllers offer robustness to such uncertainties but often lack precision on terrains with sparse steppable areas. Hybrid methods achieve enhanced robustness on sparse terrains by combining both methods but are computationally demanding and constrained by the inherent limitations of model-based planners. To achieve generalized legged locomotion on diverse terrains while preserving the robustness of learning-based controllers, this paper proposes an attention-based map encoding conditioned on robot proprioception, which is trained as part of the controller using reinforcement learning. We show that the network learns to focus on steppable areas for future footholds when the robot dynamically navigates diverse and challenging terrains. We synthesized behaviors that exhibited robustness against uncertainties while enabling precise and agile traversal of sparse terrains. In addition, our method offers a way to interpret the topographical perception of a neural network. We have trained two controllers for a 12-degrees-of-freedom quadrupedal robot and a 23-degrees-of-freedom humanoid robot and tested the resulting controllers in the real world under various challenging indoor and outdoor scenarios, including ones unseen during training.<\/jats:p>","DOI":"10.1126\/scirobotics.adv3604","type":"journal-article","created":{"date-parts":[[2025,8,27]],"date-time":"2025-08-27T17:58:14Z","timestamp":1756317494000},"update-policy":"https:\/\/doi.org\/10.34133\/aaas_crossmark","source":"Crossref","is-referenced-by-count":4,"title":["Attention-based map encoding for learning generalized legged locomotion"],"prefix":"10.1126","volume":"10","author":[{"ORCID":"https:\/\/orcid.org\/0009-0008-5928-1237","authenticated-orcid":true,"given":"Junzhe","family":"He","sequence":"first","affiliation":[{"name":"Robotic Systems Lab, ETH Zurich, 8092 Zurich, Switzerland."}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4173-154X","authenticated-orcid":true,"given":"Chong","family":"Zhang","sequence":"additional","affiliation":[{"name":"Robotic Systems Lab, ETH Zurich, 8092 Zurich, Switzerland."}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1851-2875","authenticated-orcid":true,"given":"Fabian","family":"Jenelten","sequence":"additional","affiliation":[{"name":"Robotic Systems Lab, ETH Zurich, 8092 Zurich, Switzerland."}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8971-6843","authenticated-orcid":true,"given":"Ruben","family":"Grandia","sequence":"additional","affiliation":[{"name":"Disney Research Zurich, Stampfenbachstrasse 48, 8006 Zurich, Switzerland."}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1952-1266","authenticated-orcid":true,"given":"Moritz","family":"B\u00e4cher","sequence":"additional","affiliation":[{"name":"Disney Research Zurich, Stampfenbachstrasse 48, 8006 Zurich, Switzerland."}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4285-4990","authenticated-orcid":true,"given":"Marco","family":"Hutter","sequence":"additional","affiliation":[{"name":"Robotic Systems Lab, ETH Zurich, 8092 Zurich, Switzerland."}]}],"member":"221","reference":[{"key":"e_1_3_2_2_2","doi-asserted-by":"publisher","DOI":"10.1126\/scirobotics.adh5401"},{"key":"e_1_3_2_3_2","doi-asserted-by":"publisher","DOI":"10.1126\/scirobotics.abc5986"},{"key":"e_1_3_2_4_2","doi-asserted-by":"publisher","DOI":"10.1126\/scirobotics.abk2822"},{"key":"e_1_3_2_5_2","doi-asserted-by":"publisher","DOI":"10.1126\/scirobotics.aau5872"},{"key":"e_1_3_2_6_2","doi-asserted-by":"crossref","unstructured":"J. Siekmann K. Green J. Warila A. Fern J. W. Hurst \u201cBlind bipedal stair traversal via sim-to-real reinforcement learning\u201d in Robotics: Science and Systems XVII D. A. Shell M. Toussaint M. A. Hsieh Eds. (RSS Foundation 2021); 10.15607\/RSS.2021.XVII.061.","DOI":"10.15607\/RSS.2021.XVII.061"},{"key":"e_1_3_2_7_2","unstructured":"N. Rudin D. Hoeller P. Reist M. Hutter \u201cLearning to walk in minutes using massively parallel deep reinforcement learning\u201d in Proceedings of the 5th Conference on Robot Learning A. Faust D. Hsu G. Neumann Eds. vol. 164 of Proceedings of Machine Learning Research (PMLR 2022) pp. 91\u2013100."},{"key":"e_1_3_2_8_2","doi-asserted-by":"crossref","unstructured":"N. Rudin D. Hoeller M. Bjelonic M. Hutter \u201cAdvanced skills by learning locomotion and local navigation end-to-end\u201d in 2022 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE 2022) pp. 2497\u20132503.","DOI":"10.1109\/IROS47612.2022.9981198"},{"key":"e_1_3_2_9_2","doi-asserted-by":"publisher","DOI":"10.1126\/scirobotics.adi7566"},{"key":"e_1_3_2_10_2","unstructured":"Z. Zhuang Z. Fu J. Wang C. Atkeson S. Schwertfeger C. Finn H. Zhao \u201cRobot parkour learning\u201d in Proceedings of the 7th Conference on Robot Learning (CoRL) J. Tan M. Toussaint K. Darvish Eds. vol. 229 of Proceedings of Machine Learning Research (PMLR 2023) pp. 73\u201392."},{"key":"e_1_3_2_11_2","doi-asserted-by":"crossref","unstructured":"X. Cheng K. Shi A. Agarwal D. Pathak \u201cExtreme parkour with legged robots\u201d in 2024 IEEE International Conference on Robotics and Automation (ICRA) (IEEE 2024) pp. 11443\u201311450.","DOI":"10.1109\/ICRA57147.2024.10610200"},{"key":"e_1_3_2_12_2","unstructured":"Z. Zhuang S. Yao H. Zhao \u201cHumanoid parkour learning\u201d in Proceedings of the 8th Conference on Robot Learning P. Agrawal O. Kroemer W. Burgard Eds. vol. 270 of Proceedings of Machine Learning Research (PMLR 2024) pp. 1975\u20131991."},{"key":"e_1_3_2_13_2","unstructured":"C. Zhang W. Xiao T. He G. Shi \u201cWoCoCo: Learning whole-body humanoid control with sequential contacts\u201d in Proceedings of the 8th Conference on Robot Learning P. Agrawal O. Kroemer W. Burgard Eds. vol. 270 of Proceedings of Machine Learning Research (PMLR 2024) pp. 455\u2013472."},{"key":"e_1_3_2_14_2","doi-asserted-by":"crossref","unstructured":"H. Duan B. Pandit M. S. Gadde B. Van Marum J. Dao C. Kim A. Fern \u201cLearning vision-based bipedal locomotion for challenging terrain\u201d in 2024 IEEE International Conference on Robotics and Automation (ICRA) (IEEE 2024) pp. 56\u201362.","DOI":"10.1109\/ICRA57147.2024.10611621"},{"key":"e_1_3_2_15_2","doi-asserted-by":"crossref","unstructured":"C. Zhang N. Rudin D. Hoeller M. Hutter \u201cLearning agile locomotion on risky terrains\u201d in 2024 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE 2024).","DOI":"10.1109\/IROS58592.2024.10801909"},{"key":"e_1_3_2_16_2","doi-asserted-by":"crossref","unstructured":"H. Duan A. Malik M. S. Gadde J. Dao A. Fern J. W. Hurst \u201cLearning dynamic bipedal walking across stepping stones\u201d in 2022 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE 2022) pp. 6746\u20136752.","DOI":"10.1109\/IROS47612.2022.9981884"},{"key":"e_1_3_2_17_2","doi-asserted-by":"publisher","DOI":"10.1177\/0278364910388315"},{"key":"e_1_3_2_18_2","doi-asserted-by":"crossref","unstructured":"P. Fankhauser M. Bjelonic C. Dario Bellicoso T. Miki M. Hutter \u201cRobust rough-terrain locomotion with a quadrupedal robot\u201d in 2018 IEEE International Conference on Robotics and Automation (ICRA) (IEEE 2018) pp. 5761\u20135768; 10.1109\/ICRA.2018.8460731.","DOI":"10.1109\/ICRA.2018.8460731"},{"key":"e_1_3_2_19_2","doi-asserted-by":"crossref","unstructured":"D. Dimitrov A. Sherikov P.-B. Wieber \u201cA sparse model predictive control formulation for walking motion generation\u201d in 2011 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IEEE 2011) pp. 2292\u20132299; 10.1109\/IROS.2011.6095035.","DOI":"10.1109\/IROS.2011.6095035"},{"key":"e_1_3_2_20_2","doi-asserted-by":"crossref","unstructured":"M. Neunert C. de Crousaz F. Furrer M. Kamel F. Farshidian R. Y. Siegwart J. Buchli \u201cFast nonlinear model predictive control for unified trajectory optimization and tracking\u201d in 2016 IEEE International Conference on Robotics and Automation (ICRA) (IEEE 2016) pp. 1398\u20131404.","DOI":"10.1109\/ICRA.2016.7487274"},{"key":"e_1_3_2_21_2","doi-asserted-by":"crossref","unstructured":"F. Farshidian E. Jelavic A. Satapathy M. Giftthaler J. Buchli \u201cReal-time motion planning of legged robots: A model predictive control approach\u201d in 2017 IEEE-RAS 17th International Conference on Humanoid Robotics (Humanoids) (IEEE 2017) pp. 577\u2013584.","DOI":"10.1109\/HUMANOIDS.2017.8246930"},{"key":"e_1_3_2_22_2","doi-asserted-by":"crossref","unstructured":"J. Di Carlo P. M. Wensing B. Katz G. Bledt S. Kim \u201cDynamic locomotion in the MIT Cheetah 3 through convex model-predictive control\u201d in 2018 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE 2018) pp. 1\u20139; 10.1109\/IROS.2018.8594448.","DOI":"10.1109\/IROS.2018.8594448"},{"key":"e_1_3_2_23_2","doi-asserted-by":"crossref","unstructured":"R. Grandia F. Farshidian R. Ranftl M. Hutter \u201cFeedback MPC for torque-controlled legged robots\u201d in 2019 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IEEE 2019) pp. 4730\u20134737; 10.1109\/IROS40897.2019.8968251.","DOI":"10.1109\/IROS40897.2019.8968251"},{"key":"e_1_3_2_24_2","doi-asserted-by":"crossref","unstructured":"C. Mastalli W. Merkt G. Xin J. Shim M. Mistry I. Havoutis S. Vijayakumar Agile maneuvers in legged robots: A predictive control approach. arXiv:2203.07554 [cs.RO] (2022).","DOI":"10.21203\/rs.3.rs-1870369\/v1"},{"key":"e_1_3_2_25_2","doi-asserted-by":"publisher","DOI":"10.1109\/TRO.2023.3275384"},{"key":"e_1_3_2_26_2","doi-asserted-by":"publisher","DOI":"10.1109\/TRO.2022.3186804"},{"key":"e_1_3_2_27_2","doi-asserted-by":"publisher","DOI":"10.1177\/02783649241273645"},{"key":"e_1_3_2_28_2","doi-asserted-by":"crossref","unstructured":"V. Dh\u2019Edin A. K. Chinnakkonda Ravi A. Jordana H. Zhu A. Meduri L. Righetti B. Sch\u00f6lkopf M. Khadiv \u201cDiffusion-based learning of contact plans for agile locomotion\u201d in 2024 IEEE-RAS 23rd International Conference on Humanoid Robots (Humanoids) (IEEE 2024) pp. 637\u2013644; 10.1109\/Humanoids58906.2024.10769875.","DOI":"10.1109\/Humanoids58906.2024.10769875"},{"key":"e_1_3_2_29_2","doi-asserted-by":"publisher","DOI":"10.1145\/3386569.3392432"},{"key":"e_1_3_2_30_2","doi-asserted-by":"crossref","unstructured":"O. Melon M. Geisert D. Surovik I. Havoutis M. Fallon \u201cReliable trajectories for dynamic quadrupeds using analytical costs and learned initializations\u201d in 2020 IEEE International Conference on Robotics and Automation (ICRA) (IEEE 2020) pp. 1410\u20131416; 10.1109\/ICRA40945.2020.9196562.","DOI":"10.1109\/ICRA40945.2020.9196562"},{"key":"e_1_3_2_31_2","unstructured":"D. Surovik O. Melon M. Geisert M. Fallon I. Havoutis \u201cLearning an expert skill-space for replanning dynamic quadruped locomotion over obstacles\u201d in Proceedings of the 2020 Conference on Robot Learning J. Kober F. Ramos C. Tomlin Eds. vol. 155 of Proceedings of Machine Learning Research (PMLR 2021) pp. 1509\u20131518."},{"key":"e_1_3_2_32_2","doi-asserted-by":"crossref","unstructured":"O. Melon R. Orsolino D. Surovik M. Geisert I. Havoutis M. Fallon \u201cReceding-horizon perceptive trajectory optimization for dynamic legged locomotion with learned initialization\u201d in 2021 IEEE International Conference on Robotics and Automation (ICRA) (IEEE 2021) pp. 9805\u20139811; 10.1109\/ICRA48506.2021.9560794.","DOI":"10.1109\/ICRA48506.2021.9560794"},{"key":"e_1_3_2_33_2","doi-asserted-by":"publisher","DOI":"10.1109\/TRO.2022.3172469"},{"key":"e_1_3_2_34_2","doi-asserted-by":"crossref","unstructured":"Z. Xie X. Da B. Babich A. Garg M. v. de Panne \u201cGLiDE: Generalizable quadrupedal locomotion in diverse environments with a centroidal model\u201d in Algorithmic Foundations of Robotics XV S. M. LaValle J. M. O\u2019Kane M. Otte D. Sadigh P. Tokekar Eds. (Springer International Publishing 2023) pp. 523\u2013539.","DOI":"10.1007\/978-3-031-21090-7_31"},{"key":"e_1_3_2_35_2","doi-asserted-by":"crossref","unstructured":"T. Miki L. Wellhausen R. Grandia F. Jenelten T. Homberger M. Hutter \u201cElevation mapping for locomotion and navigation using GPU\u201d in 2022 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE 2022) pp. 2273\u20132280.","DOI":"10.1109\/IROS47612.2022.9981507"},{"key":"e_1_3_2_36_2","unstructured":"A. Vaswani N. Shazeer N. Parmar J. Uszkoreit L. Jones A. N. Gomez L. u. Kaiser I. Polosukhin \u201cAttention is all you need\u201d in Advances in Neural Information Processing Systems I. Guyon U. V. Luxburg S. Bengio H. Wallach R. Fergus S. Vishwanathan R. Garnett Eds. (Curran Associates 2017) vol. 30; https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2017\/file\/3f5ee243547dee91fbd053c1c4a845aa-Paper.pdf."},{"key":"e_1_3_2_37_2","unstructured":"R. Yang M. Zhang N. Hansen H. Xu X. Wang \u201cLearning vision-guided quadrupedal locomotion end-to-end with cross-modal transformers\u201d in The 10th International Conference on Learning Representations (ICLR) (ICLR 2022); https:\/\/openreview.net\/forum?id=nhnJ3oo6AB."},{"key":"e_1_3_2_38_2","doi-asserted-by":"publisher","DOI":"10.1126\/scirobotics.adi9579"},{"key":"e_1_3_2_39_2","doi-asserted-by":"crossref","unstructured":"M. Hutter C. Gehring D. Jud A. Lauber C. D. Bellicoso V. Tsounis J. Hwangbo K. Bodie P. Fankhauser M. Bloesch R. Diethelm S. Bachmann A. Melzer M. Hoepflinger \u201cANYmal - a highly mobile and dynamic quadrupedal robot\u201d in 2016 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE 2016) pp. 38\u201344; 10.1109\/IROS.2016.7758092.","DOI":"10.1109\/IROS.2016.7758092"},{"key":"e_1_3_2_40_2","unstructured":"Fourier Intelligence Fourier GR1 (2024); https:\/\/www.fftai.com\/products-gr1."},{"key":"e_1_3_2_41_2","doi-asserted-by":"crossref","unstructured":"M. Chignoli D. Kim E. Stanger-Jones S. Kim \u201cThe MIT humanoid robot: Design motion planning and control for acrobatic behaviors\u201d in 2020 IEEE-RAS 20th International Conference on Humanoid Robots (Humanoids) (IEEE 2021) pp. 1\u20138.","DOI":"10.1109\/HUMANOIDS47582.2021.9555782"},{"key":"e_1_3_2_42_2","doi-asserted-by":"publisher","DOI":"10.1109\/TRO.2022.3222958"},{"key":"e_1_3_2_43_2","unstructured":"A. Dosovitskiy L. Beyer A. Kolesnikov D. Weissenborn X. Zhai T. Unterthiner M. Dehghani M. Minderer G. Heigold S. Gelly J. Uszkoreit N. Houlsby \u201cAn image is worth 16x16 words: Transformers for image recognition at scale\u201d in The 9th International Conference on Learning Representations (ICLR) (ICLR 2021); https:\/\/openreview.net\/forum?id=YicbFdNTTy."},{"key":"e_1_3_2_44_2","doi-asserted-by":"crossref","unstructured":"T. Miki J. Lee L. Wellhausen M. Hutter \u201cLearning to walk in confined spaces using 3D representation\u201d in 2024 IEEE International Conference on Robotics and Automation (ICRA) (IEEE 2024) pp. 8649\u20138656; 10.1109\/ICRA57147.2024.10610271.","DOI":"10.1109\/ICRA57147.2024.10610271"},{"key":"e_1_3_2_45_2","doi-asserted-by":"crossref","unstructured":"E. Chane-Sane P.-A. Leziart T. Flayols O. Stasse P. Sou\u00e8res N. Mansard \u201cCaT: Constraints as terminations for legged locomotion reinforcement learning\u201d in 2024 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE 2024) pp. 13303\u201313310; 10.1109\/IROS58592.2024.10802334.","DOI":"10.1109\/IROS58592.2024.10802334"},{"key":"e_1_3_2_46_2","doi-asserted-by":"crossref","unstructured":"J. Lee L. Schroth V. Klemm M. Bjelonic A. Reske M. Hutter \u201cExploring constrained reinforcement learning algorithms for quadrupedal locomotion\u201d in 2024 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE 2024) pp. 11132\u201311138; 10.1109\/IROS58592.2024.10801341.","DOI":"10.1109\/IROS58592.2024.10801341"},{"key":"e_1_3_2_47_2","doi-asserted-by":"publisher","DOI":"10.1109\/TRO.2024.3400935"},{"key":"e_1_3_2_48_2","unstructured":"J. Schulman F. Wolski P. Dhariwal A. Radford O. Klimov Proximal policy optimization algorithms. arXiv:1707.06347 [cs.LG] (2017)."},{"key":"e_1_3_2_49_2","unstructured":"Qualisys; https:\/\/www.qualisys.com\/."},{"key":"e_1_3_2_50_2","unstructured":"J. He C. Zhang F. Jenelten R. Grandia M. B\u00e4cher M. Hutter Data for \u201cAttention-based map encoding for learning generalized legged locomotion \u201d Zenodo (2024); 10.5281\/zenodo.14499786."}],"container-title":["Science Robotics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.science.org\/doi\/pdf\/10.1126\/scirobotics.adv3604","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,8,27]],"date-time":"2025-08-27T17:58:23Z","timestamp":1756317503000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.science.org\/doi\/10.1126\/scirobotics.adv3604"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,8,27]]},"references-count":49,"journal-issue":{"issue":"105","published-print":{"date-parts":[[2025,8,27]]}},"alternative-id":["10.1126\/scirobotics.adv3604"],"URL":"https:\/\/doi.org\/10.1126\/scirobotics.adv3604","relation":{},"ISSN":["2470-9476"],"issn-type":[{"value":"2470-9476","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,8,27]]},"assertion":[{"value":"2024-12-16","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-07-29","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-08-27","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"eadv3604"}}