BeyondMimic: From motion tracking to versatile humanoid control via guided diffusion
Combines agile motion tracking with guided diffusion to compose humanoid skills for previously unseen objectives.
SCIENCE ROBOTICS · 2024–2026
The robot learning index.
Research, organized by what robots do and how they learn.
68 papers / Updated Sep 9, 2026 ON THE COVER
6 papers featured on the journal cover
All papers 68
Newest firstRobot in a crib: How a playing robot helps us understand sensorimotor contingency learning
Uses prediction and curiosity in an embodied humanoid to study how action–sensation relationships emerge.
Evolution of humanoid locomotion control
Reviews the evolution of humanoid locomotion from model-based control to learning and generative approaches.
Learning vision-driven reactive soccer skills for humanoid robots
Trains an onboard vision-driven soccer controller that remains reactive under perceptual noise and detection failures.
SONIC: Supersizing motion tracking for natural humanoid whole-body control
Scales motion-tracking data, model capacity, and compute into a generalist humanoid whole-body controller.
PaperProjectCode
Details & sources
Zhengyi Luo, Ye Yuan, Tingwu Wang, Chenran Li, Fernando Castañeda, Sirui Chen, Zi-Ang Cao, Jiefeng Li, David Minor, Qingwei Ben, Jinhyung Park, David Sami, Zi Wang, Xingye Da, Runyu Ding, Cyrus Hogg, Lina Song, Edy Lim, Eugene Jeong, Tairan He, Haoru Xue, Wenli Xiao, Simon Yuen, Jan Kautz, Yan Chang, Umar Iqbal, Linxi Jim Fan, Yuke Zhu
Science Robotics 11(117) · 10.1126/scirobotics.aed4592
Cover status: not yet verified
Source 1 Source 2 Source 3 Source 4ZEST: Zero-shot embodied skill transfer for athletic robot control
Transfers athletic motions from capture, video, and animation to different robot bodies without per-skill tuning.
PaperOpen version
Details & sources
Jean Pierre Sleiman, He Li, Alphonsus Adu-Bredu, Robin Deits, Arun Kumar, Kevin Bergamin, Mohak Bhardwaj, Scott Biddlestone, Nicola Burger, Matthew A Estrada, Francesco Iacobelli, Twan Koolen, Alexander Lambert, Erica Lin, M Eva Mungai, Zach Nobles, Shane Rozen-Levy, Yuyao Shi, Jiashun Wang, Jakob Welner, Fangzhou Yu, Mike Zhang, Alfred Rizzi, Jessica Hodgins, Sylvain Bertrand, Yeuhi Abe, Scott Kuindersma, Farbod Farshidian
Science Robotics 11(117) · 10.1126/scirobotics.aec7695
Cover status: not yet verified
Source 1 Source 2 Source 3Advances, challenges, and opportunities for legged robots
Reviews capabilities and open challenges in legged hardware, autonomy, data, and deployment.
Ultra-wideband radar to measure in vivo muscle forces
Combines radar measurements with learned estimators of muscle force for assistive sensing applications.
Bioinspired multimodal robotics
Reviews multimodal robot design and the transition toward learned planning and motion control.
Performant robotic manipulation with real-world reinforcement learning
Improves diffusion visuomotor policies through offline and online reinforcement learning, with fast one-step execution.
Agile perceptive multiskill locomotion for quadrupedal robots in the wild
Pretrains locomotion skills from optimized motions and learns perceptive transitions across challenging outdoor terrain.
PaperProject
Details & sources
Jun-Gill Kang, Jaehyun Park, Tae-Gyu Song, Joon-Ha Kim, Seungwoo Hong, Hae-Won Park
Science Robotics 11(116) · 10.1126/scirobotics.adz7397
Cover status: confirmed. KAIST reproduces the July 2026 journal cover and identifies the featured study and DOI.
Source 1 Source 2 Source 3 Source 4 Source 5 Source 6Precise aggressive aerial maneuvers with sensorimotor policies
Learns vision-to-control policies for aggressive gap traversal, using planner initialization and simulation distillation.
RAPTOR: A foundation policy for quadrotor control
Distills many specialized teachers into a recurrent quadrotor policy that adapts to unseen hardware.
A retrieval-augmented framework enabling VLM spatial awareness for object-centric robot manipulation
Grounds vision-language reasoning in object-centric geometry to execute spatially precise manipulation instructions.
From autonomy to alliance: Robotic foundation models must learn with us, not just for us
Proposes collaborative robot foundation models that account for partners, social roles, and changing interaction norms.
A careful examination of large behavior models for multitask dexterous manipulation
Evaluates multitask diffusion policies with controlled trials, showing benefits from pretraining scale and diversity.
PaperProject
Details & sources
Jose Barreiros, Andrew Beaulieu, Aditya Bhat, Rick Cory, Eric Cousineau, Hongkai Dai, Ching-Hsin Fang, Kunimatsu Hashimoto, Muhammad Zubair Irshad, Masha Itkina, Naveen Kuppuswamy, Kuan-Hui Lee, Katherine Liu, Dale McConachie, Ian McMahon, Haruki Nishimura, Calder Phillips-Grafflin, Charles Richter, Paarth Shah, Krishnan Srinivasan, Blake Wulfe, Chen Xu, Mengchao Zhang, Alex Alspach, Maya Angeles, Kushal Arora, Vitor Campagnolo Guizilini, Alejandro Castro, Dian Chen, Ting-Sheng Chu, Sam Creasey, Sean Curtis, Richard Denitto, Emma Dixon, Eric Dusel, Matthew Ferreira, Aimee Goncalves, Grant Gould, Damrong Guoy, Swati Gupta, Xuchen Han, Kyle Hatch, Brendan Hathaway, Allison Henry, Hillel Hochsztein, Phoebe Horgan, Shun Iwase, Donovon Jackson, Siddharth Karamcheti, Sedrick Keh, Joseph Masterjohn, Masayuki Masuda, Jean Mercat, Patrick Miller, Paul Mitiguy, Tony Nguyen, Jeremy Nimmer, Yuki Noguchi, Reko Ong, Aykut Onol, Owen Pfannenstiehl, Richard Poyner, Leticia Priebe Mendes Rocha, Gordon Richardson, Christopher Rodriguez, Derick Seale, Michael Sherman, Mariah Smith-Jones, David Tago, Pavel Tokmakov, Matthew Tran, Basile Van Hoorick, Igor Vasiljevic, Sergey Zakharov, Mark Zolotas, Rareș Ambruș, Kerri Fetzer-Borelli, Benjamin Burchfiel, Hadas Kress-Gazit, Siyuan Feng, Stacie Ford, Russ Tedrake
Science Robotics 11(113) · 10.1126/scirobotics.aea6201
Cover status: confirmed. The research team links its April 2026 journal cover; the institute confirms the cover feature.
Source 1 Source 2 Source 3 Source 4 Source 5 Source 6Demonstrate once, execute on many: Kinematic intelligence for cross-robot skill transfer
Embeds kinematic constraints in demonstration-derived dynamical policies for skill transfer across robot morphologies.
Milliwatt ultrasound for navigation in visually degraded environments on palm-sized aerial robots
Uses learned ultrasound denoising to navigate small aerial robots through fog, darkness, and clutter.
Cross-robot behavior adaptation through intention alignment
Aligns language-annotated intentions to transfer demonstrated behavior across heterogeneous robots and environments.
Visual-tactile pretraining and online multitask learning for humanlike manipulation dexterity
Combines human visual–tactile pretraining with online multitask learning for multifingered manipulation.
Learning realistic lip motions for humanoid face robots
Learns audio-conditioned lip trajectories for an expressive robotic face without hand-coded mouth movements.
PaperProjectOpen version
Details & sources
Yuhang Hu, Jiong Lin, Judah Allen Goldfeder, Philippe M Wyder, Yifeng Cao, Steven Tian, Yunzhe Wang, Jingran Wang, Mengmeng Wang, Jie Zeng, Cameron Mehlman, Yingke Wang, Delin Zeng, Boyuan Chen, Hod Lipson
Science Robotics 11(110) · 10.1126/scirobotics.adx3017
Cover status: confirmed. Columbia reproduces the January 2026 journal cover and explicitly identifies the featured study.
Source 1 Source 2 Source 3 Source 4 Source 5 Source 6 Source 7Deep learning-based autonomous retinal vein cannulation in ex vivo porcine eyes
Uses learned motion and event predictors to automate retinal vein cannulation in ex vivo porcine eyes.
Resilient odometry via hierarchical adaptation
Integrates learned inertial odometry with hierarchical sensor adaptation for navigation under severe sensory degradation.
Deep domain adaptation eliminates costly data required for task-agnostic wearable robotic control
Transfers accessible movement data into device-specific joint-moment estimators for task-agnostic exoskeleton assistance.
Metamaterial robotics
Reviews metamaterial robotics, including AI-supported design, modeling, and adaptive control.
Learning a thousand tasks in a day
Separates alignment and interaction, using trajectory retrieval to teach a robot many tasks from few demonstrations.
PaperProjectCodeOpen version
Details & sources
Kamil Dreczkowski, Pietro Vitiello, Vitalis Vosylius, Edward Johns
Science Robotics 10(108) · 10.1126/scirobotics.adv7594
Cover status: confirmed. The author project page reproduces the journal cover and identifies the November 2025 issue.
Source 1 Source 2 Source 3 Source 4 Source 5 Source 6 Source 7 Source 8Accelerating discovery in natural science laboratories with AI and robotics: Perspectives and challenges
Examines autonomy, reproducibility, and human collaboration challenges in AI-enabled laboratory robotics.
Paper
Details & sources
Andrew I Cooper, Patrick Courtney, Kourosh Darvish, Moritz Eckhoff, Hatem Fakhruldeen, Andrea Gabrielli, Animesh Garg, Sami Haddadin, Kanako Harada, Jason Hein, Maria Hübner, Dennis Knobbe, Gabriella Pizzuto, Florian Shkurti, Ruja Shrestha, Kerstin Thurow, Rafael Vescovi, Birgit Vogel-Heuser, Ádám Wolf, Naruki Yoshikawa, Yan Zeng, Zhengxue Zhou, Henning Zwirnmann
Science Robotics 10(106) · 10.1126/scirobotics.adv7932
Cover status: not yet verified
Source 1 Source 2Extending robot minds through collective learning
Argues for distributed robot learning through modular, interdependent systems rather than monolithic models.
A review of learning-based dynamics models for robotic manipulation
Reviews learned interaction dynamics, state representations, uncertainty, and their use in manipulation planning.
RoboBallet: Planning for multirobot reaching with graph neural networks and reinforcement learning
Learns graph-based coordination that jointly allocates tasks, schedules actions, and plans collision-free multiarm motion.
"Data will solve robotics and automation: True or false?": A debate
Presents a debate on whether scaling data alone can solve robotics and automation.
Attention-based map encoding for learning generalized legged locomotion
Uses proprioception-conditioned map attention to learn robust, precise locomotion across sparse footholds.
Learning contact-rich whole-body manipulation with example-guided reinforcement learning
Learns contact-rich whole-body object handling from example motions using a compliant humanoid upper body.
Precise and dexterous robotic manipulation via human-in-the-loop reinforcement learning
Combines demonstrations and human corrections with sample-efficient real-world learning for precise manipulation.
AI in therapeutic and assistive exoskeletons and exosuits: Influences on performance and autonomy
Reviews how machine learning supports intention recognition, assessment, and autonomous wearable assistance.
AI search, physician removal: Bronchoscopy robot bridges collaboration in foreign body aspiration
Combines learned perception and structured exploration for autonomous bronchial foreign-body search with remote physician removal.
The grand challenges of learning medical robot autonomy
Identifies barriers to learning the perception, decision-making, and action needed for medical robot autonomy.
Will your next surgeon be a robot? Autonomy and AI in robotic surgery
Reviews the transition from surgeon-operated systems toward AI-enabled autonomous surgical robots.
Surgical embodied intelligence for generalized task autonomy in laparoscopic robot-assisted surgery
Provides a surgical simulation and learning pipeline that transfers trained task policies to real surgical robots.
Paper
Details & sources
Yonghao Long, Anran Lin, Derek Hang Chun Kwok, Lin Zhang, Zhenya Yang, Kejian Shi, Lei Song, Jiawei Fu, Hongbin Lin, Wang Wei, Kai Chen, Xiangyu Chu, Yang Hu, Hon Chi Yip, Philip Wai Yan Chiu, Peter Kazanzides, Russell H Taylor, Yunhui Liu, Zihan Chen, Zerui Wang, Samuel Kwok Wai Au, Qi Dou
Science Robotics 10(104) · 10.1126/scirobotics.adt3093
Cover status: not yet verified
Source 1 Source 2SRT-H: A hierarchical framework for autonomous surgery via language-conditioned imitation learning
Pairs a language-level planner with learned low-level trajectories to execute and recover during surgical steps.
PaperProjectOpen version
Details & sources
Ji Woong Brian Kim, Juo-Tung Chen, Pascal Hansen, Lucy Xiaoyang Shi, Antony Goldenberg, Samuel Schmidgall, Paul Maria Scheikl, Anton Deguet, Brandon M White, De Ru Tsai, Richard Jaepyeong Cha, Jeffrey Jopling, Chelsea Finn, Axel Krieger
Science Robotics 10(104) · 10.1126/scirobotics.adt5254
Cover status: confirmed. Publisher announcement identifies the July 2025 surgical cover; the AAAS issue notification identifies SRT-H.
Source 1 Source 2 Source 3 Source 4 Source 5 Source 6 Source 7The robot will see you now: Foundation models are the path forward for autonomous robotic surgery
Examines the data requirements and prospects of foundation models for autonomous robotic surgery.
A compact neuromorphic system for ultra-energy-efficient, on-device robot localization
Combines event sensing and spiking networks for compact, low-energy visual place recognition on a robot.
Hierarchically depicting vehicle trajectory with stability in complex environments
Combines learned path proposals with stable trajectory optimization to navigate complex environments efficiently.
Multimodal information structuring with single-layer soft skins and high-density electrical impedance tomography
Structures redundant electrical measurements from a soft skin into multimodal touch and environmental estimates.
High-speed control and navigation for quadrupedal robots on complex and discrete terrain
Combines optimized foothold planning with a learned tracker and an adaptive training terrain distribution.
Learning coordinated badminton skills for legged manipulators
Learns coordinated whole-body badminton skills while accounting for real-world visual perception uncertainty.
Bridging hard and soft: Mechanical metamaterials enable rigid torque transmission in soft robots
Pairs torque-transmitting metamaterials with learned inverse kinematics for forceful soft-arm manipulation.
Bioinspired design of a tissue-engineered ray with machine learning
Uses machine-learning-directed geometry optimization to improve a tissue-engineered swimming robot.
The case against machine vision for the control of wearable robotics: Challenges for commercial adoption
Discusses practical barriers to deploying vision-based control in commercial wearable robots.
A neuromechanics solution for adjustable robot compliance and accuracy
Combines a muscle-inspired impedance model and adaptive cerebellar network to tune robot compliance and accuracy.
Development of compositionality through interactive learning of language and action of robots
Integrates vision, proprioception, and language to learn compositional action–language representations.
Counterfactual rewards promote collective transport using individually controlled swarm microrobots
Uses counterfactual rewards to assign credit when independently controlled microrobots collectively transport cargo.
NeuralFeels with neural fields: Visuotactile perception for in-hand manipulation
Fuses touch, vision, and proprioception to reconstruct and track previously unseen objects during in-hand manipulation.
PaperProjectCodeOpen version
Details & sources
Sudharshan Suresh, Haozhi Qi, Tingfan Wu, Taosha Fan, Luis Pineda, Mike Lambeta, Jitendra Malik, Mrinal Kalakrishnan, Roberto Calandra, Michael Kaess, Joseph Ortiz, Mustafa Mukadam
Science Robotics 9(96) · 10.1126/scirobotics.adl0628
Cover status: confirmed. CMU reproduces the journal cover and identifies NeuralFeels as the November 2024 cover.
Source 1 Source 2 Source 3 Source 4 Source 5 Source 6 Source 7 Source 8Reinforcement learning-based framework for whale rendezvous via autonomous sensing robots
Couples multiagent routing with radio-based sensing to improve autonomous rendezvous with sperm whales.
Autonomous tracking of honey bee behaviors over long-term periods with cooperating robots
Combines autonomous robotic observation with behavioral analysis to track a honey-bee queen and colony interactions.
Paper
Details & sources
Jiří Ulrich, Martin Stefanec, Fatemeh Rekabi-Bana, Laurenz Alexander Fedotoff, Tomáš Rouček, Bilal Yağız Gündeğer, Mahmood Saadat, Jan Blaha, Jiří Janota, Daniel Nicolas Hofstadler, Kristina Žampachů, Erhan Ege Keyvan, Babür Erdem, Erol Şahin, Hande Alemdar, Ali Emre Turgut, Farshad Arvin, Thomas Schmickl, Tomáš Krajník
Science Robotics 9(95) · 10.1126/scirobotics.adn6848
Cover status: not yet verified
Source 1 Source 2Intrinsic sense of touch for intuitive physical human-robot interaction
Recognizes touch trajectories on a robot body as symbolic instructions using manifold learning and neural networks.
SimPLE, a visuotactile method learned in simulation to precisely pick, localize, regrasp, and place objects
Learns visuotactile pose estimation in simulation and combines it with grasp and regrasp planning for precise placement.
Optimality principles in spacecraft neural guidance and control
Reviews neural guidance and control networks that learn optimal actions for spacecraft and drone tasks.
Wearable robots for the real world need vision
Discusses how visual information can support wearable robots in real-world environments.
Fully neuromorphic vision and control for autonomous drone flight
Trains an event-based spiking vision-to-control pipeline that transfers from simulation to onboard drone flight.
Stereoscopic artificial compound eyes for spatiotemporal perception in three-dimensional space
Combines stereoscopic artificial compound eyes with edge learning for efficient three-dimensional object tracking.
Learning robust autonomous navigation and locomotion for wheeled-legged robots
Integrates learned wheeled-legged locomotion and hierarchical navigation for kilometer-scale urban missions.
Real-world humanoid locomotion with reinforcement learning
Trains a transformer locomotion controller that adapts from observation–action history during real-world humanoid walking.
Learning agile soccer skills for a bipedal robot with deep reinforcement learning
Learns agile bipedal soccer skills and tactical behaviors in simulation for zero-shot deployment on small humanoids.
Paper
Details & sources
Tuomas Haarnoja, Ben Moran, Guy Lever, Sandy H Huang, Dhruva Tirumala, Jan Humplik, Markus Wulfmeier, Saran Tunyasuvunakool, Noah Y Siegel, Roland Hafner, Michael Bloesch, Kristian Hartikainen, Arunkumar Byravan, Leonard Hasenclever, Yuval Tassa, Fereshteh Sadeghi, Nathan Batchelor, Federico Casarini, Stefano Saliceti, Charles Game, Neil Sreendra, Kushal Patel, Marlon Gwira, Andrea Huber, Nicole Hurley, Francesco Nori, Raia Hadsell, Nicolas Heess
Science Robotics 9(89) · 10.1126/scirobotics.adi8022
Cover status: not yet verified
Source 1 Source 2Human-robot facial coexpression
Learns to anticipate human smiles and control a robotic facial model for simultaneous coexpression.
Estimating human joint moments unifies exoskeleton control, reducing user effort
Uses learned joint-moment estimates to adapt exoskeleton assistance across many movement conditions.
ANYmal parkour: Learning agile navigation for quadrupedal robots
Combines learned perception, motor skills, and high-level selection to traverse parkour-like obstacles.
DTC: Deep Tracking Control
Trains a robust tracking policy around optimized reference motions to improve precision on sparse footholds.