paper:doi-10-1126-science-1133687Resilient Machines Through Continuous Self-Modeling
Original abstract (expand)
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.
Related work— refs + corpus + external arXiv
Cited / in-corpus / arXiv badges show which signals surfaced each row. Multi-source rows weighted higher.
- Self-adapting Robotic Agents through Online Continual Reinforcement Learning with World Model FeedbackGeorg Schildbach Fabian Domberg2026≈ 73%
- Self-Consistent Model-based Adaptation for Visual Reinforcement LearningChengyang Ying, Yao Feng, Hang Su, Jun Zhu Xinning Zhou2025≈ 72%
- Learning 3D Persistent Embodied World ModelsYilun Du, Yuncong Yang, Lei Han, Peihao Chen, Dit-Yan Yeung, Chuang Gan Siyuan Zhou2025≈ 70%
- Learning State Representations via Retracing in Reinforcement LearningDong Li, Jianye Hao, Jun Wang, Neil Burgess Changmin Yu2022≈ 70%
- Learning Policies for Continuous Control via Transition ModelsSerge Thill, Marcel van Gerven, Pablo Lanillos Justus Huebotter2022≈ 70%
- ≈ 70%
- Robotic World Model: A Neural Network Simulator for Robust Policy Optimization in RoboticsAndreas Krause, Marco Hutter Chenhao Li2025≈ 70%
- ResWM: Residual-Action World Model for Visual RLGabriel Adineera, Jinzhou Tan, and Jinoh Kim Jseen Zhang2026≈ 69%
- A predictive processing model of perception and action for self-other distinctionStefan Kopp Sebastian Kahl2018≈ 69%
- Modeling motor control in continuous-time Active Inference: a surveyFederico Maggiore, Antonella Maselli, Francesco Donnarumma, Domenico Maisto, Francesco Mannella, Ivilin Peev Stoianov and Giovanni Pezzulo Matteo Priorelli2024≈ 69%
- When Models Judge Themselves: Unsupervised Self-Evolution for Multimodal ReasoningKai Shi, Chuanrui Zhang, Zirui Liao, Jun Yang, Ni Yang, Qiuying Peng, Luyuan Zhang, Hangrui Xu, Tianhuang Su, Zhenyu Yang, Haonan Lu, Haoqian Wang Zhengxian Wu2026≈ 69%
- ≈ 69%
- stable-worldmodel-v1: Reproducible World Modeling Research and EvaluationQuentin Le Lidec, Dan Haramati, Nassim Massaudi, Damien Scieur, Yann LeCun, Randall Balestriero Lucas Maes2026≈ 69%
- A Machine Consciousness architecture based on Deep Learning and Gaussian ProcessesMart\'in Molina Eduardo C. Garrido Merch\'an2020≈ 69%
- The Effectiveness of World Models for Continual Reinforcement LearningMateusz Ostaszewski, Micha{\l} Bortkiewicz, Mateusz \.Zarski, Maciej Wo{\l}czyk, Jack Parker-Holder, Stephen J. Roberts and Piotr Mi{\l}o\'s Samuel Kessler2023≈ 69%
- ≈ 68%
- Learning without neurons in physical systemsin corpus2022≈ 67%
- Relating transformers to models and neural representations of the hippocampal formationin corpus2021≈ 66%
- ≈ 66%
- Model Alignment Searchin corpus2025≈ 66%
- ≈ 66%
- ≈ 66%
- ≈ 65%
- ≈ 65%
- ≈ 65%
- Collective intelligence: A unifying concept for integrating biology across scales and substratesin corpus2024≈ 65%
- ≈ 65%
- ≈ 65%
- Persona-Model Collapse in Emergent Misalignmentin corpus2026≈ 65%
Similar preprints — Semantic Scholar
Cited by (6)
- Developmental Bioelectricity: the cognitive glue enabling evolutionary scaling from physiology to mind
Developmental bioelectricity—the network of ion channels, gap junctions, and neurotransmitter-mediated Vmem (membrane potential) dynamics operating across all body cells, not just neurons—functions as
- Self-Improvising Memory: A Perspective on Memories as Agential, Dynamically Reinterpreting Cognitive Glue
Memory's function is not fidelity but salience preservation through active reinterpretation—a claim Levin defends by unifying phenomena from caterpillar-to-butterfly metamorphosis (where training-indu
- Endless forms most beautiful 2.0: teleonomy and the bioengineering of chimaeric and synthetic organisms
Cellular collectives exhibit goal-directed competency that is substrate-independent, composition-independent, and origin-independent — a property Clawson and Levin term teleonomy — and this invariant,
- Technological Approach to Mind Everywhere: An Experimentally-Grounded Framework for Understanding Diverse Bodies and Minds
TAME—Technological Approach to Mind Everywhere—formalizes a non-binary, empirically grounded framework for recognizing, comparing, and manipulating cognition across radically diverse substrates, from
- Living Things Are Not (20th Century) Machines: Updating Mechanism Metaphors in Light of the Modern Science of Machine Behavior
Bongard and Levin argue that the longstanding debate over whether living things are machines has been conducted against a 20th-century, static definition of 'machine' that modern engineering has alrea
- The computational boundary of a 'self': developmental bioelectricity drives multicellularity and scale-free cognition
Scale-Free Cognition, the framework introduced here, proposes that any coherent Self is demarcated by a 'cognitive light cone'—a spatio-temporal boundary of events a system can measure, model, and att