Robots that can adapt like animals
Open accessThe DOI matches its Crossref record; 39 of the 40 references it cites were found in a registry or an index.
Abstract
Robots have transformed many industries, most notably manufacturing, and have the power to deliver tremendous benefits to society, such as in search and rescue, disaster response, health care and transportation. They are also invaluable tools for scientific exploration in environments inaccessible to humans, from distant planets to deep oceans. A major obstacle to their widespread adoption in more complex environments outside factories is their fragility. Whereas animals can quickly adapt to injuries, current robots cannot 'think outside the box' to find a compensatory behaviour when they are damaged: they are limited to their pre-specified self-sensing abilities, can diagnose only anticipated failure modes, and require a pre-programmed contingency plan for every type of potential damage, an impracticality for complex robots. A promising approach to reducing robot fragility involves having robots learn appropriate behaviours in response to damage, but current techniques are slow even with small, constrained search spaces. Here we introduce an intelligent trial-and-error algorithm that allows robots to adapt to damage in less than two minutes in large search spaces without requiring self-diagnosis or pre-specified contingency plans. Before the robot is deployed, it uses a novel technique to create a detailed map of the space of high-performing behaviours. This map represents the robot's prior knowledge about what behaviours it can perform and their value. When the robot is damaged, it uses this prior knowledge to guide a trial-and-error learning algorithm that conducts intelligent experiments to rapidly discover a behaviour that compensates for the damage. Experiments reveal successful adaptations for a legged robot injured in five different ways, including damaged, broken, and missing legs, and for a robotic arm with joints broken in 14 different ways. This new algorithm will enable more robust, effective, autonomous robots, and may shed light on the principles that animals use to adapt to injury.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
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?
The cited source doesn’t support the claim as written.
In the citing paper
“We render images using the MuJoCo Physics Engine's [45] built-in renderer. This renderer is not intended to be photo-realistic, and physically plausible choices of textures and lighting are not needed.”
From Domain randomization for transferring deep neural networks from simulation to the real world (2017) · cites this paper as [45]
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“under Quality-Diversity optimization (Pugh et al., 2016), the search space is divided into behavioral niches to evolve diverse solutions, a paradigm that challenged the dominant approach of training a single agent, as it proved more robust in real-world settings (Cully et al., 2015).”
From Dynamics of niche construction in adaptable populations evolving in diverse environments (2023) · cites this paper as Cully et al., 2015
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Similarly, extensive reliance on simulation and offline training further distances learning from the full variability and constraint structure of real sensorimotor interaction (Jakobi et al., 1995; Cully et al., 2015).”
From Toward Enactive Artificial Intelligence (2026) · cites this paper as Cully et al., 2015
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Gaussian Processes for Machine Learning
Rasmussen, C. E., & Williams, C. K. I. (2005) · doi:10.7551/mitpress/3206.001.0001 · Cited by 11,705
Corke, P., Roberts, J., Cunningham, J., & Hainsworth, D. (2008) · Springer Handbook of Robotics · doi:10.1007/978-3-540-30301-5_50 · Cited by 3,543
A survey of robot learning from demonstration
Argall, B. D., Chernova, S., Veloso, M., & Browning, B. (2009) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2008.10.024 · Cited by 3,227
Reinforcement learning in robotics: A survey
Kober, J., Bagnell, J. A., & Peters, J. (2013) · The International Journal of Robotics Research · doi:10.1177/0278364913495721 · Cited by 3,160 · cited 9×
“The key insight is that, whereas current learning algorithms either start with no knowledge of the search space 17 or with minimal knowledge from a few human demonstrations 17,18 , animals better understand the space of possible behaviors and their value from previous experience 19 , enabling…”
BackgroundMethodResultInfluentialDiagnosis and Fault-Tolerant Control
Blanke, M., Kinnaert, M., Lunze, J., & Staroswiecki, M. (2003) · Springer eBooks · doi:10.1007/978-3-662-05344-7 · Cited by 2,789
Nonlinear Programming: Sequential Unconstrained Minimization Techniques
Fiacco, A. V., & McCormick, G. P. (1968) · Defense Technical Information Center (DTIC) · Cited by 2,262
Stanley: The robot that won the DARPA Grand Challenge
Thrun, S., Montemerlo, M., Dahlkamp, H., Stavens, D., Aron, A., Diebel, J., Fong, P., Gale, J., Halpenny, M., Hoffmann, G., et al. (2006) · Journal of Field Robotics · doi:10.1002/rob.20147 · Cited by 2,237 · cited 2×
“3C, undam-aged: 24 [16; 41] seconds, 3 [2; 5] physical trials, repaired: 29 [16; 82] seconds, 3.5 [2; 10] trials).”
BackgroundInfluentialBrochu, E., Cora, V. M., & de Freitas, N. (2010) · arXiv (Cornell University) · doi:10.48550/arxiv.1012.2599 · Cited by 2,011
Bayesian integration in sensorimotor learning
Körding, K. P., & Wolpert, D. M. (2004) · Nature · doi:10.1038/nature02169 · Cited by 2,004 · cited 1×
“…optimization procedure followed by our robot appears similar to the technique employed by humans when they optimize an unknown function 23 , and there is strong evidence that animal brains learn probability distributions, combine them with prior knowledge, and act as Bayesian optimizers 25,26 .”
MethodControl of mental activities by internal models in the cerebellum
Ito, M. (2008) · Nature Reviews Neuroscience · doi:10.1038/nrn2332 · Cited by 1,356 · cited 1×
“A final parallel is that the simulator and Gaussian process components of Intelligent Trial and Error are two forms of predictive models, which are known to exist in animals 29,12 .”
Research works printed but not linked: 2
- Siciliano, B. & Khatib, O. Springer Handbook of Robotics (Springer, 2008). · doi:10.1007/978-3-540-30301-5
- Kluger, J. & Lovell, J. Apollo 13 (Mariner Books, 2006).
Cited by
Items 401–500 of 1,007
Xie, Z., & Jin, Y. (2018) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2018.2796940 · Cited by 16
“, the optimization of certain function [11], [12], [14], statistically optimal prediction on future system states [4] or optimal surviving robustness [6], [7].”
BackgroundAn approach to evolve and exploit repertoires of general robot behaviours
Gomes, J., Oliveira, S. M., & Christensen, A. L. (2018) · Swarm and Evolutionary Computation · doi:10.1016/j.swevo.2018.06.009 · Cited by 19
“[14], for instance, study how a repertoire of hexapod locomotion behaviours can be used to make the robot fault tolerant: when the robot is impaired (for example, a broken leg), the controller can look for new locomotion behaviours that can cope with that fault.”
BackgroundBianconi, F., & Filippucci, M. (2019) · Lecture notes in civil engineering · doi:10.1007/978-3-030-03676-8_1 · Cited by 21
Swarm robotics: Robustness, scalability, and self-X features in industrial applications
Heinrich, M. K., Soorati, M. D., Kaiser, T. K., Wahby, M., & Hamann, H. (2019) · it - Information Technology · doi:10.1515/itit-2019-0003 · Cited by 17
“There are for instance hexapod robots that adapt like animals after damage to some of their legs [8], and here we describe our method for structures that self-heal using a model inspired by plants.”
MethodDeshpande, A. M., Kumar, R., Minai, A. A., & Kumar, M. (2020) · ASME Dynamic Systems and Control Conference (DSCC) · doi:10.1115/dscc2020-3319 · Cited by 16
Gaining Insight into Quality Diversity
Auerbach, J. E., Iacca, G., & Floreano, D. (2016) · Genetic and Evolutionary Computation Conference Companion (GECCO Companion) · doi:10.1145/2908961.2931675 · Cited by 16
“If this is the case, then it seems likely that ability of evolution to produce a diversity of solutions (e.g. self models [2], features [1, 10], motor actions [3]) may be what justifies their existence.”
BackgroundExplore, Discover and Learn: Unsupervised Discovery of State-Covering Skills
Campos, V., Trott, A., Xiong, C., Socher, R., Giró-i-Nieto, X., & Torres, J. (2020) · UPCommons institutional repository (Universitat Politècnica de Catalunya) · doi:10.48550/arxiv.2002.03647 · Cited by 126
“Quality Diversity algorithms aim at combining the best of both worlds by optimizing task-specific fitness functions while encouraging diverse behaviors in a population of agents (Cully et al., 2015; Mouret & Clune, 2015; Pugh et al., 2016).”
BackgroundMoldtmann, P., Berk, J., Ryan, S., Klavzar, A., Limido, J., Lange, C., Rana, S., & Venkatesh, S. (2024) · Defence Technology · doi:10.1016/j.dt.2024.05.007 · Cited by 15
“The utility of BO has been demonstrated across a wide range of applications, traditionally in machine learning [20] but more recently in applied science and engineering domains such as robotics [21], material discovery [22], and material constitutive modelling [23].”
Robust Humanoid Locomotion Using Trajectory Optimization and Sample-Efficient Learning
Yeganegi, M. H., Khadiv, M., Moosavian, S. A. A., Zhu, J.-J., Prete, A. D., & Righetti, L. (2019) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids43949.2019.9035003 · Cited by 15
“locomotion of a quadruped parameterized by a walk engine [14], locomotion of a hexapod [15], balancing of an inverted pendulum [16], finding task priority in inverse dynamics for controlling dual-arm manipulators [17], and scheduling contact for a one-leg hopper [18].”
BackgroundExobiology Extant Life Surveyor (EELS)
Carpenter, K., Thoesen, A., Mick, D., Martia, J., Cable, M. L., Mitchell, K. L., Hovsepian, S., Jasper, J., Georgiev, N., Thakker, R., et al. (2021) · ASCE International Conference on Engineering, Science, Construction, and Operations in Challenging Environments (Earth and Space) · doi:10.1061/9780784483374.033 · Cited by 17
Multi-Time Resolution Ensemble LSTMs for Enhanced Feature Extraction in High-Rate Time Series
Barzegar, V., Laflamme, S., Hu, C. K., & Dodson, J. C. (2021) · Sensors · doi:10.3390/s21061954 · Cited by 16
“Such knowledge integration can be done, for example, in the form of accompanying logic rules [28], algebraic equations [29], and mecha-Sensors nistic models [30,31].”
BackgroundMulti-Level Evolution for Robotic Design
Chand, S., & Howard, D. (2021) · Frontiers in Robotics and AI · doi:10.3389/frobt.2021.684304 · Cited by 18
“This algorithm has been successfully used for tackling problems with 2 – 6 feature dimensions (Cully et al., 2015).”
MethodWang, B., Cui, X., Sun, J., & Gao, Y. (2021) · International Journal of Advanced Robotic Systems · doi:10.1177/17298814211044934 · Cited by 15
“Even with a failed leg, the hexapod robot can go on walking.(4) The traditional hexapod robot motion law is determined through its kinematics and dynamics modeling.”
BackgroundFrom Adaptive Locomotion to Predictive Action Selection – Cognitive Control for a Six-Legged Walker
Schilling, M., Paskarbeit, J., Ritter, H. J., Schneider, A., & Cruse, H. (2021) · IEEE Transactions on Robotics · doi:10.1109/tro.2021.3106832 · Cited by 17
Multiplexed Chemical Sensing CMOS Imager
Wang, D., Qian, L., Zhang, F., Mallires, K. R., Tipparaju, V. V., Yu, J., Forzani, E., Jia, C., Yang, Q., Tao, N. J., et al. (2022) · ACS Sensors · doi:10.1021/acssensors.2c01277 · Cited by 15
Nitschke, G., & Didi, S. (2017) · Frontiers in Robotics and AI · doi:10.3389/frobt.2017.00062 · Cited by 15
“, 2015) has provided increasing empirical evidence that maintaining diversity in phenotypes (robot behaviors) improves the quality (task performance) of evolved behaviors (Mouret and Doncieux, 2012; Cully et al., 2015; Cully and Mouret, 2016; Gomes et al., 2016).”
BackgroundInfluentialOhtake, T. (2018) · Molecular Systems Design & Engineering · doi:10.1039/c8me00020d · Cited by 16
Li, D., Wu, W., & Qiu, Z. (2023) · Actuators · doi:10.3390/act12040157 · Cited by 14
Tung, Y.-S., Luebbers, M. B., Roncone, A., & Hayes, B. (2024) · ACM/IEEE International Conference on Human-Robot Interaction (HRI) · doi:10.1145/3610977.3635003 · Cited by 14
“Inspired by the success of QD algorithms in fnding diverse solutions in large search spaces [4, 24], we use MAP-Elites to search for environments that best elicit legible human behavior.”
BackgroundMethodFlow currents support simple and versatile trail-tracking strategies
Hang, H., Jiao, Y., Merel, J., & Kanso, E. A. (2026) · Physical Review Research · doi:10.1103/qj9q-p7rg · Cited by 15
Huang, P., Ding, W., Stoler, B., Francis, J., Chen, B., & Zhao, D. X. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11127319 · Cited by 24
Intrinsically Motivated Exploration of Learned Goal Spaces
Laversanne-Finot, A., Péré, A., & Oudeyer, P. (2021) · Frontiers in Neurorobotics · doi:10.3389/fnbot.2020.555271 · Cited by 17
“Discovering multiple ways to manipulate an object or perform a certain skill can also help quickly repair strategies in case of damages (Cully et al., 2015).”
BackgroundDeforMoBot: A Bio-Inspired Deformable Mobile Robot for Navigation Among Obstacles
Mulvey, B. W., Lalitharatne, T. D., & Nanayakkara, T. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3273393 · Cited by 15
“A trial-and-error learning algorithm was introduced in [19] to conduct experiments that allow robots to rapidly discover a behavior that adapts to damage.”
BackgroundA testbed that evolves hexapod controllers in hardware
Heijnen, H., Howard, D., & Kottege, N. (2017) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2017.7989128 · Cited by 15
“Recent evolutionary efforts have focused on using a simulator, together with a limited number of hardware evaluations in a co-evolutionary context [13], [14], however they have a focus on system identification and controller recovery following a partial hardware failure, rather than generating mission-specific optimal…”
BackgroundThe Evolutionary Buffet Method
Hintze, A., Schossau, J., & Bohm, C. (2018) · Genetic Programming Theory and Practice · doi:10.1007/978-3-030-04735-1_2 · Cited by 18
Kano, T., Kanauchi, D., Ono, T., Aonuma, H., & Ishiguro, A. (2019) · Frontiers in Neurorobotics · doi:10.3389/fnbot.2019.00104 · Cited by 15
“, 2018) and trial-and-error methods (Cully et al., 2015), however, the performance level of robots using these techniques is not satisfactory.”
BackgroundBehavioral Repertoires for Soft Tensegrity Robots
Doney, K., Petridou, A., Karaul, J., Khan, A. F., Liu, G., & Rieffel, J. A. (2020) · IEEE Symposium Series on Computational Intelligence (SSCI) · doi:10.1109/ssci47803.2020.9308218 · Cited by 14
Computational Psychiatry for Computers
Schulz, E., & Dayan, P. (2020) · iScience · doi:10.1016/j.isci.2020.101772 · Cited by 14
“, 2015), as well as building robots that can cope with damage (Cully et al., 2015) is an exciting topic of ongoing research.”
BackgroundEvolving the Behavior of Machines: From Micro to Macroevolution
Mouret, J.-B. (2020) · iScience · doi:10.1016/j.isci.2020.101731 · Cited by 19
“ng the many situations for which it is useful to generate thousands of high-quality solutions. One of the most prominent illustrations of MAP-Elites is to generate gait repertoires for legged robots (Cully et al., 2015; Pautrat et al., 2018)(Figure 2D). Cully et al. (2015)thus leveraged MAP-Elites to create about 15,0…”
InfluentialSelf-citationEnsemble feature extraction for multi-container quality-diversity algorithms
Cazenille, L. (2021) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3449639.3459392 · Cited by 17
Bayesian Optimization in Variational Latent Spaces with Dynamic CompressionPreprint
Antonova, R., Rai, A., Li, T., & Kragić, D. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1907.04796 · Cited by 26
“For example [20] use simulation and user-defined features to transform the space of a 36-dimensional controller into 6D, making the search for walking controllers of a hexapod much more data-efficient.”
MethodLi, J., Yang, C., Ding, L., You, B., Li, W., Zhang, X., & Gao, H. (2024) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2024.3410522 · Cited by 13
“P ECENTLY, multilegged walking robots have been widely studied in increasing domains, such as exploring and transporting items in the field or planetary environments, largely owing to their good loading stability and strong terrain trafficability [1], [2].”
Expressivity of parameterized and data-driven representations in quality diversity search
Hagg, A., Berns, S., Asteroth, A., Colton, S., & Bäck, T. (2021) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3449639.3459287 · Cited by 13
“The QD algorithm used in this work is based on an alternative formulation to MAP-Elites [8].”
BackgroundMethodPI-ARS: Accelerating Evolution-Learned Visual-Locomotion with Predictive Information Representations
Lee, K.-H., Nachum, O., Zhang, T., Guadarrama, S., Tan, J., & Yu, W. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981952 · Cited by 19
“Evolution Strategy (ES) optimization techniques have received increasing interest in recent years within the robotics and deep reinforcement learning (DRL) communities [1], [2], [3], [4], [5], [6], [7].”
BackgroundSampling diverse near-optimal solutions via algorithmic quantum annealing
Mohseni, M., Rams, M. M., Isakov, S. V., Eppens, D., Pielawa, S., Strümpfer, J., Boixo, S., & Neven, H. (2023) · Physical review. E · doi:10.1103/physreve.108.065303 · Cited by 17
“The quality diversity has applications in adaptive robotic systems that could be inherently equipped with a diverse set of high-performing agents or policies to effectively cope with uncertainties, damages, and changing environmental conditions [13].”
BackgroundWoodford, G. W., du Plessis, M. C., & Pretorius, C. J. (2016) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2016.11.018 · Cited by 16
Evaluating MAP-Elites on constrained optimization problems
Fioravanzo, S., & Iacca, G. (2019) · Genetic and Evolutionary Computation Conference Companion (GECCO Companion) · doi:10.1145/3319619.3321939 · Cited by 14
“riven by novelty [13, 18, 24] or diversity [4, 20], rather than the objective alone, is a relatively new area of research in the Evolutionary Computation community. Among these algorithms, MAP-Elites [3, 19] has attracted quite some attention in the field, due to its simplicity and general applicability. Since its int…”
BackgroundMethodInfluentialA Framework for Automatic Behavior Generation in Multi-Function Swarms
Engebråten, S. A., Moen, J., Yakimenko, O. A., & Glette, K. (2020) · Frontiers in Robotics and AI · doi:10.3389/frobt.2020.579403 · Cited by 14
“Evolving a large repertoire of controllers or behaviors before the robot is deployed might allow it to recover from otherwise crippling physical and hardware faults (Cully et al., 2015; Mouret and Clune, 2015; Cully and Mouret, 2016).”
BackgroundInfluentialQuality and Diversity in Evolutionary Modular Robotics
Nordmoen, J., Veenstra, F., Ellefsen, K. O., & Glette, K. (2020) · IEEE Symposium Series on Computational Intelligence (SSCI) · doi:10.1109/ssci47803.2020.9308511 · Cited by 15
“The repertoire should aid in overcoming premature convergence in addition to serving as a store that can be utilized at design or runtime to select the best morphology for the given task [15], [16].”
AutoFac: The Perpetual Robot Machine
Nitschke, G., & Howard, D. (2021) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2021.3104789 · Cited by 16
Human-level design proposals by an artificial agent in multiple scenarios
Serra, G., & Miralles, D. (2021) · Design Studies · doi:10.1016/j.destud.2021.101029 · Cited by 14
Modular Robotic Limbs for Astronaut Activities Assistance
Zhao, S., Zhao, J., Sui, D., Wang, T., Zheng, T., Zhao, C., & Zhu, Y. (2021) · Sensors · doi:10.3390/s21186305 · Cited by 15
Ning, M., Shao, L., Chen, F., Li, M., Zhang, C., & Zhang, Q. (2019) · Mathematical Problems in Engineering · doi:10.1155/2019/8261617 · Cited by 13
“For example, if the three module legs are installed adjacent to each other, the robot cannot achieve good motion performance.”
BackgroundFast Adaptation via Policy-Dynamics Value FunctionsPreprint
Răileanu, R., Goldstein, M., Szlam, A. D., & Fergus, R. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.02879 · Cited by 14
“A popular approach for fast adaptation to new environments is meta reinforcement learning (meta RL) (Cully et al., 2015; Finn et al., 2017; Wang et al., 2017; Duan et al., 2016; Xu et al., 2018; Houthooft et al., 2018; Sæmundsson et al., 2018; Nagabandi et al., 2018; Humplik et al., 2019; Rakelly et al., 2019).”
BackgroundChattunyakit, S., Kobayashi, Y., Emaru, T., & Ravankar, A. A. (2019) · Applied Sciences · doi:10.3390/app9040799 · Cited by 18
Interactive Trajectory Adaptation through Force-guided Bayesian Optimization
Rozo, L. D. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8968571 · Cited by 12
“BayesOpt has recently gained interest in different robotic applications, such as behavior adaptation for damaged robots [8], automatic controller tuning for balancing [9], locomotion [10], and interaction tasks [11], and in physical HRC [12], [13].”
BackgroundMachine Behaviour (Originally Published 2019 by Springer Nature)
Rahwan, I., Cebrián, M., Obradovich, N., Bongard, J., Bonnefon, J., Breazeal, C., Crandall, J. W., Christakis, N. A., Couzin, I. D., Jackson, M. O., et al. (2022) · Wiley eBooks · doi:10.1002/9781119815075.ch10 · Cited by 13
Bayesian Optimization with Ensemble Learning Models and Adaptive Expected Improvement
Polyzos, K. D., Lu, Q., & Giannakis, G. B. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10095008 · Cited by 12
“Emails: { polyz003,qlu,georgios } @umn.edu tuning [1], robotics [2, 3], sensor networks [4], and drug discovery [5] to name a few.”
BackgroundThe Role of Morphological Variation in Evolutionary Robotics: Maximizing Performance and Robustness
Carvalho, J. T., & Nolfi, S. (2023) · Evolutionary Computation · doi:10.1162/evco_a_00336 · Cited by 16
“…is necessary to select robots that are robust to environmental variations (Branke, 2012; Bredeche et al., 2012; Milano et al., 2019; Pagliuca and Nolfi, 2019) and which can eventually adapt to varying conditions on the fly (Floreano and Urzelai, 2000; Risi and Stanley, 2012; Cully et al., 2015).”
BackgroundDiscovering Adaptable Symbolic Algorithms from Scratch
Kelly, S., Park, D., Song, X., McIntire, M., Nashikkar, P., Guha, R., Banzhaf, W., Deb, K., Boddeti, V. N., Tan, J., et al. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10341979 · Cited by 12
“Unlike previous studies in learning quadruped locomotion in the presence of non-stationary morphologies (e.g., [9]), we are specifically interested in how controllers can be automatically built from scratch without requiring any prior task decomposition or curriculum learning.”
BackgroundEvolving Robust Robot Team Morphologies for Collective Construction
Watson, J., & Nitschke, G. (2015) · IEEE Symposium Series on Computational Intelligence (SSCI) · doi:10.1109/ssci.2015.150 · Cited by 16
“Also, robots may be damaged as they interact with their environment [28], [29], in which case evolved behaviors must continue to operate in modified robot bodies (morphologies).”
BackgroundA Platform That Directly Evolves Multirotor Controllers
Howard, D. (2017) · IEEE Transactions on Evolutionary Computation · doi:10.1109/tevc.2017.2703142 · Cited by 12
“robot [17].”
MABE (Modular Agent Based Evolver): A framework for digital evolution research
Bohm, C., G., N. C., & Hintze, A. (2017) · European Conference on Artificial Life · doi:10.1162/isal_a_016 · Cited by 12
Learning Kinematic Structure Correspondences Using Multi-Order Similarities
Chang, H. J., Fischer, T., Petit, M., Zambelli, M., & Demiris, Y. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2777486 · Cited by 14
Task-Agnostic Evolution of Diverse Repertoires of Swarm Behaviours
Gomes, J., & Christensen, A. L. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-00533-7_18 · Cited by 17
Cazenille, L., Bredèche, N., & Aubert-Kato, N. (2019) · IEEE Symposium Series on Computational Intelligence (SSCI) · doi:10.1109/ssci44817.2019.9003047 · Cited by 12
“This method is particularly suited to cope with multi-modal, deceptive, hard exploration problems where traditional optimization algorithms would be prone to premature convergence, as in evolutionary robotics [19], [9], [11].”
Hatchuel, A., Masson, P. L., Thomas, M., & Weil, B. (2021) · Proceedings of the Design Society · doi:10.1017/pds.2021.603 · Cited by 13
“, 2016, Lehman and Stanley, 2011b) or to design images (Woolley and Stanley, 2011); - Map-Eliteswas used to design a self-repairing hexapod robot (Cully et al., 2015) or wing profile (Cully et al.”
BackgroundMethodInfluentialQuality-Diversity Meta-Evolution: Customizing Behavior Spaces to a Meta-Objective
Bossens, D. M., & Tarapore, D. (2022) · IEEE Transactions on Evolutionary Computation · doi:10.1109/tevc.2022.3152384 · Cited by 12
“In robotics, this includes the design of robot morphologies and controllers [12], [14], as well as providing the behavior space for behavior adaptation algorithms [15], [16], which search for high-performing behaviors in the evolved archive to help robots recover rapidly from environmental changes or damages to their…”
Self-citationPSTO: Learning Energy-Efficient Locomotion for Quadruped Robots
Zhu, W., & Rosendo, A. (2022) · Machines · doi:10.3390/machines10030185 · Cited by 12
“To conduct massive parameter tuning, population optimization include black-box [26] and Bayesian optimization [27] is an appealing approach.”
MethodWang, J., Pan, X., Luo, W., Shuai, Y., Xie, Q., Xu, J., Song, Z., Wu, C., & Zhang, W. (2022) · Advanced Electronic Materials · doi:10.1002/aelm.202201064 · Cited by 12
A Complexity Framework for Self-Engineering Systems
Brooks, S., & Roy, R. (2020) · Smart and Sustainable Manufacturing Systems · doi:10.1520/ssms20200059 · Cited by 12
Kume, A., Matsumoto, E., Takahashi, K., Ko, W., & Tan, J. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1710.06117 · Cited by 13
“[4], which is based on evolutionary strategies, to quickly adapt to abnormalities.”
BackgroundMethodInfluentialA survey on control of humanoid fall over
Subburaman, R., Kanoulas, D., Tsagarakis, N. G., & Lee, J. (2023) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2023.104443 · Cited by 11
“An impressive work on this type of robot is introduced in [88], where hexapod robots learn how to continue locomoting after they damage or lose any of their legs.”
BackgroundAnne, T., & Mouret, J.-B. (2024) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3638529.3653993 · Cited by 11
“MAP-Elites[44]hasshownpromising resultsinroboticswithevolvingarepertoireofdiversegaitsfor leggedrobots[11],orinvideo-gamingindustriestogeneratepro-cedural contents like spaceships and game levels [25].”
Self-citationRobustness to Leg Loss in Opiliones: A Review and Framework Considerations for Future Research
Escalante, I., & O’Brien, S. L. (2024) · Integrative and Comparative Biology · doi:10.1093/icb/icae051 · Cited by 12
Mulvey, B. W., & Nanayakkara, T. (2024) · Scientific Reports · doi:10.1038/s41598-024-75607-7 · Cited by 11
Comparing multimodal optimization and illumination
Vassiliades, V., Chatzilygeroudis, K. I., & Mouret, J.-B. (2017) · Genetic and Evolutionary Computation Conference Companion (GECCO Companion) · doi:10.1145/3067695.3075610 · Cited by 11
“The Multi-dimensional Archive of Phenotypic Elites (MAP-Elites) [1, 8] algorithm offers a different solution: it discretizes the behavior space into a number of niches, storing in every niche only the elite individual throughout the evolutionary simulation.”
BackgroundSelf-citationCombining STDP and binary networks for reinforcement learning from images and sparse rewards
Chevtchenko, S. F., & Ludermir, T. B. (2021) · Neural Networks · doi:10.1016/j.neunet.2021.09.010 · Cited by 11
Design Principles for Neurorobotics
Krichmar, J. L., & Hwu, T. J. (2022) · Frontiers in Neurorobotics · doi:10.3389/fnbot.2022.882518 · Cited by 12
“For example, Cully et al. (2015) developed a novel method for adapting gaits on a hexapod robot.”
MethodGe, C., Duan, Z., Li, R., Chen, H., Li, T., Hu, P., Wang, Z., Zhao, J., & Zhang, J. (2022) · Sensors and Actuators A Physical · doi:10.1016/j.sna.2022.113676 · Cited by 11
First Do Not Fall: Learning to Exploit a Wall With a Damaged Humanoid Robot
Anne, T., Dalin, E., Bergonzani, I., Ivaldi, S., & Mouret, J.-B. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3188884 · Cited by 11
Self-citationSafeAPT: Safe Simulation-to-Real Robot Learning Using Diverse Policies Learned in Simulation
Kaushik, R., Arndt, K., & Kyrki, V. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3177294 · Cited by 17
Ferigo, A., Soros, L. B., Medvet, E., & Iacca, G. (2022) · Conference on Artificial Life (ALIFE) · doi:10.1162/isal_a_00493 · Cited by 11
“In particular, we use two fundamentally different Evolutionary Algorithms (EAs): a simple form of Evolution Strategies (ES) (Beyer and Schwefel, 2002), and the Quality Diversity algorithm MAP-Elites (Cully et al., 2015).”
BackgroundMethodLiu, Y., & Farimani, A. B. (2022) · Journal of Mechanisms and Robotics · doi:10.1115/1.4055167 · Cited by 11
Keyframe-based CPG for Stable Gait Design and Online Transitions in Legged Robots
Shaw, S., & Sartoretti, G. (2022) · IEEE Conference on Decision and Control (CDC) · doi:10.1109/cdc51059.2022.9992457 · Cited by 11
“In particular, some animals are able to seamlessly transition to gaits that use a fraction of their limbs, freeing up the others for manipulation tasks or to handle limb injuries [2], [3].”
BackgroundAoi, S., Yabuuchi, Y., Morozumi, D., Okamoto, K., Adachi, M., Senda, K., & Tsuchiya, K. (2023) · Soft Robotics · doi:10.1089/soro.2022.0177 · Cited by 11
“17 18–28 Although the use of a large number legs has advantages for legged robots, it makes the body long and complicates the interaction with the environment.”
InfluentialAI-driven automated discovery tools reveal diverse behavioral competencies of biological networks
Etcheverry, M., Moulin-Frier, C., Oudeyer, P., & Levin, M. G. (2025) · eLife · doi:10.31219/osf.io/s6thq · Cited by 17
Jiang, L., Yang, F., Zhang, M., & Yang, Z. (2023) · Materials Today Communications · doi:10.1016/j.mtcomm.2023.107584 · Cited by 11
Mongeau, J.-M., Yang, Y. J., Escalante, I., Cowan, N. J., & Jayaram, K. (2024) · Integrative and Comparative Biology · doi:10.1093/icb/icae121 · Cited by 12
“…demonstrate an impressive c apacit y for motor le ar ning th at en ables ag i le maneuv ers ov er ha rsh ( Hwa ngbo et al. 2019 ; Lee et al. 2020 ) and deform able n atural terrains ( Choi et al. 2023 ; Guizzo 2019 ) and even com pensa te for leg loss ( Cu l ly et a l. 2015 ; Yang et a l. 2020 ).”
BackgroundRapidly adapting robot swarms with Swarm Map-based Bayesian Optimisation
Bossens, D. M., & Tarapore, D. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9560958 · Cited by 11
“1) that extends the Intelligent Trial and Error (IT&E) algorithm [12] by accounting for a suitable behavioural descriptor for swarms, providing interrobot communication of the selected controller and its estimated performance, and by allowing an asynchronous, decentralised Bayesian optimisation variant such that the b…”
BackgroundMethodInfluentialSelf-citationSequence-Based Selection Hyper-Heuristic Model via MAP-Elites
Sanchez, M., Cruz‐Duarte, J. M., Ortíz-Bayliss, J. C., & Amaya, I. (2021) · IEEE Access · doi:10.1109/access.2021.3106815 · Cited by 11
“This search domain is also known as behavior space [23], [28], [60], but we opted for the first name in this work.”
BackgroundOSCAR: Data-Driven Operational Space Control for Adaptive and Robust Robot Manipulation
Wong, J., Makoviychuk, V., Anandkumar, A., & Zhu, Y. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9811967 · Cited by 20
Exploring Model Inversion Attacks in the Black-box Setting
Dionysiou, A., Vassiliades, V., & Athanasopoulos, E. (2023) · Proceedings on Privacy Enhancing Technologies · doi:10.56553/popets-2023-0012 · Cited by 15
Simulation of Tactile Sensing Arrays for Physical Interaction Tasks
Kappassov, Z., Ramón, J. A. C., & Perdereau, V. (2020) · IEEE/ASME International Conference on Advanced Intelligent Mechatronics · doi:10.1109/aim43001.2020.9158822 · Cited by 11
“For example, whereas the simulation environment is sufficient to gather the data for solving the problem of learning adaptation behaviors to faults without contacts [1], real experiments are needed to collect and process raw signals for solving the problem of safe physical interactions [2].”
BackgroundQR Code-Based Self-Calibration for a Fault-Tolerant Industrial Robot Arm
Zhang, Y., Zhu, W., & Rosendo, A. (2019) · IEEE Access · doi:10.1109/access.2019.2920429 · Cited by 11
“[8] developed a hexapod robot and a manipulator that can adapt its behavior after a malfunction by using a method called Intelligent Trial and Error (IT&E) algorithm.”
BackgroundMethodInfluentialEfficient reinforcement learning control for continuum robots based on Inexplicit Prior KnowledgePreprint
Liu, J., Shou, J., Fu, Z., Zhou, H., Xie, R., Zhang, J., Fei, J., & Zhao, Y. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.11573 · Cited by 12
“Except for creating data-driven surrogate models as MBRL algorithms do, the article states that there is another way to let robots adapt with micro-data: leverage prior knowledge on the policy parameters [17] [18], on the expected return [19], or on the dynamic models [20] [21].”
MethodZhi, J., Wang, R., Clune, J., & Stanley, K. O. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.11164 · Cited by 12
“…Mnih et al. 2016; Ecoffet et al. 2019; Nair et al. 2015; Horgan et al. 2018; Jader-berg et al. 2018; Sutton and Barto 2018) and population-based methods (Salimans et al. 2017; Jaderberg et al. 2017; Such et al. 2017; Conti et al. 2018; Wang et al. 2019; Stanley et al. 2019; Cully et al. 2015).”
MethodSelf-citationDesign and Visualization of a Hierarchical Metamaterial with Tunable Stiffness
Xi, K., Jiang, X., Zhao, D., Chen, G., Ma, J., & Chen, Y. (2025) · Research · doi:10.34133/research.0874 · Cited by 10
Fault-Aware Robust Control via Adversarial Reinforcement Learning
Yang, F., Yang, C., Guo, D., Liu, H., & Sun, F. (2021) · IEEE International Conference on Cyber Technology in Automation, Control, and Intelligent Systems · doi:10.1109/cyber53097.2021.9588329 · Cited by 15
From intelligent agents to trustworthy human-centred multiagent systems
Soorati, M. D., Gerding, E., Marchioni, E., Naumov, P., Norman, T. J., Ramchurn, S. D., Rastegari, B., Sobey, A. J., Stein, S., Tarapore, D., et al. (2022) · AI Communications · doi:10.3233/aic-220127 · Cited by 10
“We have also proposed rapid adaptation of robots to quickly recover from a variety of damages [35].”
Self-citationSun, T., Dai, Z., & Manoonpong, P. (2023) · Frontiers in Neural Circuits · doi:10.3389/fncir.2023.1111285 · Cited by 10
“…al., 2017; Hwangbo et al., 2019; Ishige et al., 2019; Jones et al., 2020, ant-colony optimization Juang and Yeh, 2018, intelligent trial and error Cully et al., 2015, and black-box optimization Thor et al., 2020) are as follows: • It does not require robot kinematics, environmental models, and…”
BackgroundMethodPhysical Scaffolding Accelerates the Evolution of Robot Behavior
Buckingham, D., & Bongard, J. (2017) · Artificial Life · doi:10.1162/artl_a_00236 · Cited by 10
“Various approaches have been proposed to combat the reality gap problem, including proactive methods that increase robustness of behavior before transferral [21, 27], retroactive methods that return data obtained from reality to increase the likelihood of success during the next transferral [5, 26, 11], and extreme me…”
BackgroundBehavioral Repertoire via Generative Adversarial Policy Networks
Jegorova, M., Doncieux, S., & Hospedales, T. M. (2019) · IEEE International Conference on Development and Learning (ICDL) · doi:10.1109/devlrn.2019.8850727 · Cited by 10
Two-Stage Robot Controller Auto-Tuning Methodology for Trajectory Tracking Applications
Roveda, L., Forgione, M., & Piga, D. (2020) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2020.12.276 · Cited by 12
“In particular, (Drieß et al., 2017) employs constrained BO to select three force-controller parameters for combined position/interaction tasks; (Cully et al., 2015) describes a trial-and-error algorithm that allows robots to adapt their behaviour in presence of damage; (Yuan et al., 2019) proposes…”
BackgroundA geometric encoding for neural network evolution
Templier, P., Rachelson, E., & Wilson, D. G. (2021) · Annual Conference on Genetic and Evolutionary Computation · doi:10.1145/3449639.3459361 · Cited by 7
“Quality diversity methods such as MAP-Elites search explicitly for diversity in policy space [6].”
MethodRelevance-guided unsupervised discovery of abilities with quality-diversity algorithms
Grillotti, L., & Cully, A. (2022) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3512290.3528837 · Cited by 14
“[10], with the Duty Factor as hand-coded BD.”
BackgroundMethodInfluentialSelf-citationAutomatic Tuning and Selection of Whole-Body Controllers
D’Elia, E., Mouret, J.-B., Kober, J., & Ivaldi, S. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9981058 · Cited by 13
“[22] encode priors into the GP mean function by first precomputing a behavior-performance map, which is a map in parameter space of how well the robot performs in simulation, and then using this map as the GP mean initial”
MethodSelf-citationLexicase Selection for Multi-Task Evolutionary Robotics
Stanton, A., & Moore, J. M. (2022) · Artificial Life · doi:10.1162/artl_a_00374 · Cited by 12
“Generalizing controllers spans learning and reacting to environmental contexts across many environments (Lehman et al., 2013), adapting and reconfiguring morphology in response to damage (Kriegman et al., 2019), and exhibiting multiple gaits for one morphology (Cully et al., 2015).”
BackgroundComplex Three-Dimensional Terrains Traversal of Insect-Scale Soft Robot
Liu, Y., Liang, J., Lu, J., Chen, H., Miao, Z., Wang, D., Wang, X., & Zhang, M. (2022) · Soft Robotics · doi:10.1089/soro.2022.0073 · Cited by 10
Trial-and-error learning of repulsors for humanoid QP-based whole-body control
Spitz, J., Bouyarmane, K., Ivaldi, S., & Mouret, J.-B. (2017) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2017.8246914 · Cited by 10
“Our main insight is that even if a model makes inaccurate predictions for some behaviors, this is not necessarily the case for all the behaviors [15], [16], [7], [17].”
BackgroundMethodInfluentialSelf-citation
Metadata
- Authors
- Cully, A., Clune, J., Tarapore, D., & Mouret, J.-B.
- Affiliations
- Centre National de la Recherche Scientifique (FR)
- University of Wyoming (US)
- Corresponding author
- Mouret, Jean-Baptiste
- Venue
- Nature
- Published
- 2015-05-27
- Volume
- 521
- Issue
- 7553
- Pages
- 503-507
- ISSN
- 0028-0836 (ISSN-L)
- 1476-4687
- OpenAlex
- W1738827650
- Cited by
- 979
- References
- 40
- Citations by year
- 2026: 51 · 2025: 76 · 2024: 88 · 2023: 90 · 2022: 100 · 2021: 115 · 2020: 129 · 2019: 112 · 2018: 78 · 2017: 78 · 2016: 48 · 2015: 13
- FWCI
- 70.47
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- www.springer.com/tdm (tdm)
- www.springer.com/tdm (vor)
- Locations
- Nature (publishedVersion)
- PubMed (publishedVersion)
- ePrints Soton (University of Southampton) (acceptedVersion)
- arXiv (Cornell University) (submittedVersion)
- HAL (submittedVersion, other-oa)
- RePEc: Research Papers in Economics (submittedVersion)
- arXiv (Cornell University)
- Funders
- Agence Nationale de la Recherche (ANR-12-JS03-0009)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Physical Sciences › Engineering › Mechanical Engineering › Modular Robots and Swarm Intelligence
- Physical Sciences › Computer Science › Artificial Intelligence › Evolutionary Algorithms and Applications
- Keywords
- compensatory behavior, trial-and-error learning, autonomous mobile robots, legged robots, robot manipulators, prior knowledge, robot robustness, search and rescue robots, disaster response
- MeSH terms
- Adaptation, Physiological, Algorithms, Animals, Artificial Intelligence, Behavior, Animal, Dogs, Extremities, Motor Skills, Robotics, Time Factors, Biomimetics
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.