Resilient Machines Through Continuous Self-Modeling
The DOI matches its Crossref record; 31 of the 31 references it cites were found in a registry or an index.
Abstract
Animals sustain the ability to operate after injury by creating qualitatively different compensatory behaviors. Although such robustness would be desirable in engineered systems, most machines fail in the face of unexpected damage. We describe a robot that can recover from such change autonomously, through continuous self-modeling. A four-legged machine uses actuation-sensation relationships to indirectly infer its own structure, and it then uses this self-model to generate forward locomotion. When a leg part is removed, it adapts the self-models, leading to the generation of alternative gaits. This concept may help develop more robust machines and shed light on self-modeling in animals.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Bongard et al. (2006) have created an artificial ''starfish'' that gradually develops an explicit internal self-model.”
From Empirical perspectives from the self-model theory of subjectivity: a brief summary with examples (2007) · cites this paper as Bongard et al. (2006)
In the citing paper
“When part of its leg is removed, it adapts its self-model and generates alternative gaits -it learns to limp.”
From Empirical perspectives from the self-model theory of subjectivity: a brief summary with examples (2007) · cites this paper as Bongard et al. (2006)
In the citing paper
“Their four-legged machine uses actuation-sensation relationships to indirectly infer its own structure and then uses this self-model to generate forward locomotion.”
From Empirical perspectives from the self-model theory of subjectivity: a brief summary with examples (2007) · cites this paper as Bongard et al. (2006)
Citations
Cites
System Identification: Theory for the User
Lennart Ljung, A. (1987) · Cited by 17,690
Thrun, S. (2002) · Communications of the ACM · doi:10.1145/504729.504754 · Cited by 7,994
(1999) · Choice Reviews Online · doi:10.5860/choice.36-5109 · Cited by 2,528
(1987) · Medical Entomology and Zoology · Cited by 2,512
Efficient Bipedal Robots Based on Passive-Dynamic Walkers
Collins, S., Ruina, A., Tedrake, R., & Wisse, M. (2005) · Science · doi:10.1126/science.1107799 · Cited by 1,873
Modelling and Control of Robot Manipulators
Sciavicco, L., & Siciliano, B. (2000) · Advanced Textbooks in Control and Signal Processing · doi:10.1007/978-1-4471-0449-0 · Cited by 1,701
Introduction to Machine Learning
(2019) · Series in Machine Perception and Artificial Intelligence · doi:10.1142/9789811201967_0001 · Cited by 1,639
Evolutionary Robotics: The Biology, Intelligence, and Technology of Self-Organizing Machines
Nolfi, S., & Floreano, D. (2000) · The MIT Press eBooks · doi:10.7551/mitpress/2889.001.0001 · Cited by 1,513
Crick, F., & Koch, C. (2003) · Nature Neuroscience · doi:10.1038/nn0203-119 · Cited by 1,492
RHex: A Simple and Highly Mobile Hexapod Robot
Saranli, U., Buehler, M., & Koditschek, D. E. (2001) · The International Journal of Robotics Research · doi:10.1177/02783640122067570 · Cited by 1,322
Research works printed but not linked: 3
- H. S. Seung, M. Opper, H. Sompolinsky, in Proceedings of the 5th Workshop on Computational Learning Theory (ACM Press, New York, 1992), pp. 287–294.
- J. Bongard, H. Lipson, J. Mach. Learn. Res.6, 1651 (2005).
- R. A. Brooks, in Proceedings of the 1st European Conference on Artificial Life, F. J. Varela, P. Bourgine, Eds. (Springer-Verlag, Berlin, 1992), pp. 3–10.
Cited by
Items 801–861 of 861
Étude de croissance de créaturesartificielles résilientes
Akrour, D. (2020) · University of Biskra Theses Repository (University of Biskra)
Mechanisms of anosognosia: semiology and functional imaging.
Jedidi, H. (2013) · ORBi (University of Liège)
Vico-Vela, F. J. (2008) · Repositorio Institucional de la Universidad de Málaga (University of Málaga)
Developing Empathy in Social Robots
Graham, M. (2025) · PhilPapers (PhilPapers Foundation)
Kyambadde, I. (2023) · White Rose eTheses Online (University of Leeds, The University of Sheffield, University of York)
Apprentissage efficace en données pour la robotique à l'aide de simulateurs
Kaushik, R. (2020) · theses.fr (ABES)
Rodríguez Gómez, M. (2024) · Deposito de Investigacion Universidad de Sevilla (University of Seville)
Organoid SentiencePreprint
Verma, S. (2024) · PhilSci-Archive (University of Pittsburgh)
Biological machines and the mechanization of life
Trogemann, G. (2013) · e-Publications@khm (Academy of Media Arts Cologne)
Chaotic exploration and learning of locomotor behaviours
Shim, Y. (2013) · Sussex Research Online (University of Sussex)
Principi e metodi della robotica evolutiva
Fantini, A. (2014) · AMS Degree Thesis (University of Bologna)
Klaas, A. (2014) · Habilitation Regulations of the Faculty of Mechanical Engineering (University of Paderborn)
Brains as output/input systems
Brembs, B. (2013) · doi:10.59350/qx0gn-n8773
Yeryüzü evrim modeli benzetimlerinin doğrusal olmayan dinamiklerinin aralıklı olarak saptanması
Birol, Ö. D. (2024) · Istanbul Technical University Academic Open Archive (Istanbul Technical University)
Damage Identification and Mitigation in Multi-Legged Robotic Systems
Farghdani, S. (2025) · Carleton University's Institutional Repository (MacOdrum Library, Carleton University) · doi:10.22215/etd/2025-16796
Getting robots back on track by reconstituting control in unexpected situations with online learning
Allard, M., Flageat, M., Lim, B., & Cully, A. (2026) · Nature Communications · doi:10.1038/s41467-026-70256-y
“This sentence implies some offline hyperparameter tuning that contradicts earlier claims.”
Controversy over the existence of non-conceptual content
Kołodziejczyk, S. (2010) · Jagiellonian University Repository (Jagiellonian University)
Mechanistic Foundations of Goal-Directed ControlPreprint
Lago, A. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.15248
“Conversely, embod-ied artificial systems grow in sophistication [4], but lack the resilience and context-sensitivity of biological ones [5,6].”
Evidence of an Emergent "Self" in Continual Robot LearningPreprint
Jhunjhunwala, A., Goldfeder, J., & Lipson, H. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.24350
“5 More recent methods reconstruct the robot’s body visually, 6 discover morphology from physical interaction data, 7 or infer a body schema from exteroceptive and proprioceptive signals.”
Self-citationContributions to Simulation-based High-dimensional Sequential Decision Making
Hoock, J. (2013) · theses.fr (ABES) · doi:10.70675/0d84e4c5zd7e5z4e77z9ad0z93c0e11f8833
Exploratory Robotic Controllers : An Evolution and Information Theory Driven Approach
Zhang, G. (2015) · theses.fr (ABES) · doi:10.70675/be25a63az960fz47f4z803ezeee0a9e0d714
Embodied Intelligence at the Intersection of Science Fiction and Reality
Ross, A. G. (2026) · IOP Conference Series Materials Science and Engineering · doi:10.1088/1757-899x/1343/1/012012
La genèse de la représentation de soi : une approche néo-humienne
Charrier, A. (2021) · theses.fr (ABES) · doi:10.70675/f7bf6301za687z402czb3cdz355d6f893cc1
Learning-Based Fault Detection for Legged Robots in Remote Dynamic EnvironmentsPreprint
Stewart-Height, A., Jahagirdar, S., & Matni, N. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.03397
“Model-based optimization techniques have frequently been used to expose unanticipated faults [4, 7, 8, 9, 29], while not as common in the literature, machine learning approaches to unanticipated faults have also been proposed [14, 25].”
Multi-task Optimization over Networks of Tasks
Hatzky, J., Bartz–Beielstein, T., Eiben, A. E., & Yaman, A. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-36226-1_34
Dutta, B. J. (2025) · Artificial Intelligence and Cloud Computing Conference (AICCC) · doi:10.1145/3789982.3790046
Continuum Robot Modeling with Action Conditioned Flow Matching
Lin, J., Ruan, J., & Lipson, H. (2026) · Robotics: Science and Systems · doi:10.15607/rss.2026.xxii.198
Self-citationEmbodied Evolutionary Self-Adaptation to Damage in Multi-Legged RobotsPreprint
Farghdani, S., & Chhabra, R. (2026) · Research Square · doi:10.21203/rs.3.rs-9706251/v1
The placeholder god: psychedelic religion and the suspension of reference
Sawyer, C. (2026) · International Journal for Philosophy of Religion · doi:10.1007/s11153-026-10016-6
“Bongard et al. ( 2006) showed how robotic systems achieved resilience through continuous self-modeling: by recur - sively updating an internal image of their own structure, machines could adapt to injury and maintain functionality.”
Suvorov, V. V. (2026) · Frontiers in Robotics and AI · doi:10.3389/frobt.2026.1811996
“…representations without generative decoding (LeCun, 2022); self-modeling robotics enables agents to autonomously discover their own morphology (Bongard et al., 2006; Kwiatkowski and Lipson, 2019); and algebraic inverse-dynamics methods achieve real-time optimal control for analytically…”
Model-based Optimization of Anguilliform Swimming Gaits for Soft Robotic ApplicationsPreprint
Stratum, B. V., Gallentine, J. C., Rucker, D. C., Barth, E., Clark, J. E., & Shoele, K. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.11278
“Achieving comparable agility and robustness in robotic platforms that are employed for multiple environmental conditions necessitates efficient models that capture essential fluid–structure interactions at a low enough complexity for use in real-time design and control tasks [3].”
Adaptive simulation platform in nvidia isaac sim for self-aware heterogeneous robots
Naseri, A., Plosila, J., & Haghbayan, M.-H. (2026) · European Conference on Modelling and Simulation (ECMS) · doi:10.7148/2026-0404
Yang, C. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.12254
“Bongard, Zykov, and Lipson demonstrated that a robot could continuously infer a model of its own morphology and use that model to recover after damage [Bongard et al., 2006].”
System-Self as a Data Structure: An Architectural Approach to Bounded Adaptation
Franz, E., & Alhassan, Y. (2026) · Procedia Computer Science · doi:10.1016/j.procs.2026.08.184
Robot Learning to Communicate through Projected Visual AbstractionsPreprint
Yan, D., Wang, B., Liu, J., & Chen, B. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.22434
Learning Governing Equations for Datasets
Sahimi, M. (2026) · Wiley eBooks · doi:10.1002/9783527851935.ch16
Klimaj, V., & Safron, A. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-33010-9_28
Instinct and reflex inspired safety mechanisms for adaptive robots
Bertoni, E. (2026) · AMS Degree Thesis (University of Bologna)
Construction of Machine Self-Awareness in Embodied Intelligence
Li, Y. (2026) · IEEE China Conference on System Simulation Technology and its Applications (CCSSTA) · doi:10.1109/ccssta69471.2026.11635781
“Digital twins are sometimes described as self-models because they maintain updated representations of a physical system and simulate its possible behavior [14].”
Tian, Y. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7345456
Active Embodied Intervention Grounds Physical-Law Understanding in Artificial IntelligencePreprint
Yamaki, N., & Churiki, T. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7352878
A Proposal on Generalisable Emergent Synchronisation via Swarm‐Inspired Central Pattern Generators
Chen, J., Chen, S., Fan, L., & Xu, C. (2026) · IET Cyber-Systems and Robotics · doi:10.1049/csy2.70066
Moreddu, R., Greulich, P., Baumgartner, C., Borghi, F., & Levin, M. (2026) · Research Square · doi:10.21203/rs.3.rs-10895632/v1
Kosta, A. K., & Roy, K. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2609.25687
“Fault-tolerant locomotion has been addressed through damage-specific search, self-modeling, behavioral selection, and adaptive CPG dynamics [30], [37]–[39].”
Embodied Agentic Harnesses for Physical Intelligence A SurveyPreprint
Huang, Y., Chen, Y., Yang, Z., & Ding, G. (2026) · Engineering Archive (engrXiv) · doi:10.31224/8297 · Cited by 1
Phan, T. T. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7549499
Brilhante, D. D. S. F. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7556614
Cognitive evolution enables de novo design of untethered millimeter-scale robotsPreprint
Li, H., Zhao, W., Tao, Z., Li, P., Wang, X., Wang, Q., Wang, B., Zhai, Y., Wei, W., Wang, S., et al. (2026) · Research Square · doi:10.21203/rs.3.rs-10770716/v1
Nine Trials to Recover: A Reproducible Benchmark for Repertoire-Free Soft-Robot Damage AdaptationPreprint
Zhang, S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2610.07616
“Self-modeling robots infer changes in their embodiment and synthesize compensatory behavior [1].”
Mulani, Z. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7598779
Analytic Robot Self-Model: Learning Screw-Axis Kinematics for Rendering and Model-Based ControlPreprint
Xu, C., & Tan, N. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2610.15558
“…rely on geometry and kinematic parameters specified in advance by engineers, whereas self-modelling lets the robot obtain this information from its own sensing and motion, an attractive option when a robot is hard to model precisely, its structure is unknown or its body changes during use [1–5].”
A Sense of Proprioception in Language ModelsPreprint
Geiping, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2610.17495
“…analogues are common in control theory and robotics, given that from a cybernetics perspective, systems that integrate feedback about internal states are more robust (von Holst and Mittelstaedt, 1950; Wolpert et al., 1998; Bongard et al., 2006; Kwiatkowski and Lipson, 2019; Lee et al., 2020).”
Kim, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2610.20838
“Post-damage adaptation has advanced through self-modeling, which revises a robot’s body model after component loss [3], and Intelligent Trial and Error (IT&E), which uses a library of behaviors computed offline to guide a small number of online tests [4].”
Krakowski, S. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7641340
Physical AI: Concepts, Techniques, and ChallengesPreprint
Zhang, W., Shen, C., Xue, T., Zhao, B., Jiang, B., Guo, Z., Cui, Z., Li, Y., Zhou, M., Chen, X., et al. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2610.27045
Zhang, H., Zhou, Z., Chen, Y., Hu, L., Liu, Y., Zhou, D., Huang, Z., & Xu, H. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2610.33445
“The robot studied by Bongard, Zykov, and Lipson inferred its own structure from action and sensing, then found new ways to move after damage [15].”
Petit, O., & Velasco, C. (2026) · Oxford University Press eBooks · doi:10.1093/9780198909088.003.0009
The Making of Engineered Minds: From Accidental Emergence to Deliberate Creation
Moritz, E. (2026) · PhilPapers (PhilPapers Foundation)
Supporting information 1 of Simulation-Based Internal Models for Safer Robots
Blum, C., Winfield, A. F. T., & Hafner, V. V. (2018) · doi:10.3389/frobt.2017.00074.s001
Metadata
- Authors
- Bongard, J., Zykov, V., & Lipson, H.
- Affiliations
- Cornell University (US)
- Corresponding author
- Bongard, Josh
- Venue
- Science
- Published
- 2006-11-17
- Volume
- 314
- Issue
- 5802
- Pages
- 1118-1121
- ISSN
- 0036-8075 (ISSN-L)
- 1095-9203
- OpenAlex
- W2001685400
- Cited by
- 709
- References
- 31
- Citations by year
- 2026: 24 · 2025: 22 · 2024: 31 · 2023: 22 · 2022: 32 · 2021: 36 · 2020: 39 · 2019: 39 · 2018: 37 · 2017: 53 · 2016: 43 · 2015: 40 · 2014: 33 · 2013: 48 · 2012: 33
- FWCI
- 26.07
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Science (publishedVersion)
- PubMed (publishedVersion)
- Topics
- Physical Sciences › Engineering › Control and Systems Engineering › Robotic Locomotion and Control
- Life Sciences › Neuroscience › Cognitive Neuroscience › Embodied and Extended Cognition
- Physical Sciences › Engineering › Biomedical Engineering › Prosthetics and Rehabilitation Robotics
- Keywords
- self-modeling, four-legged robot, legged locomotion, gait adaptation, structural inference, autonomous recovery, robustness
- MeSH terms
- Algorithms, Animals, Computer Simulation, Robotics
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.