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 301–400 of 1,007
Self-organizing systems: what, how, and why?
Gershenson, C. (2025) · npj Complexity · doi:10.1038/s44260-025-00031-5 · Cited by 32
“Some have been inspired in the collective behavior of animals 118 , and their potential applications are vast.”
Directed Locomotion for Modular Robots with Evolvable Morphologies
Lan, G., Jelisavcic, M., Roijers, D. M., Haasdijk, E., & Eiben, A. E. (2018) · Lecture notes in computer science · doi:10.1007/978-3-319-99253-2_38 · Cited by 32
“Last, an alternative approach based on machine learning for adaptive locomotion was proposed by Cully et al., to account for changes in body properties [9].”
MethodThe dynamic of body and brain co-evolution
Pagliuca, P., & Nolfi, S. (2021) · Adaptive Behavior · doi:10.1177/1059712321994685 · Cited by 30
“the evolution of robots capable to recover automatically from body damage (Cully et al., 2015; Kriegman et al., 2019), and (c) the evolution of robots capable to vary their morphology to adapt to varying environmental conditions (Shah et al.”
BackgroundXie, B., & Maciejewski, A. A. (2021) · IEEE Transactions on Robotics · doi:10.1109/tro.2021.3082017 · Cited by 23
“Fault-tolerant motion planning strategies can be applied to recover from joint failures [8], [9] or be incorporated in anticipation of failures.”
BackgroundJoint Friction during Deployment of a Near-Full-Scale Tensegrity Footbridge
Sychterz, A. C., & Smith, I. F. C. (2017) · Journal of Structural Engineering · doi:10.1061/(asce)st.1943-541x.0001817 · Cited by 23
“Incorporating biomimetic behavior (mimicry of natural33 processes) such as learning and self-diagnosis has potential to enhance behavior of the struc-34 ture (Adam and Smith 2007)(Cully et al. 2015)(Lobo and Vico 2010).”
BackgroundCully, A., Chatzilygeroudis, K. I., Allocati, F., & Mouret, J.-B. (2018) · The Journal of Open Source Software · doi:10.21105/joss.00545 · Cited by 23
Self-citationApplying machine learning to accelerate new materials development
Wu, W., & Sun, Q. (2018) · Zhongguo kexue. Wulixue Lixue Tianwenxue · doi:10.1360/sspma2018-00073 · Cited by 25
QED: Using Quality-Environment-Diversity to Evolve Resilient Robot Swarms
Bossens, D. M., & Tarapore, D. (2020) · IEEE Transactions on Evolutionary Computation · doi:10.1109/tevc.2020.3036578 · Cited by 28
Self-citationEfficient Novelty Search Through Deep Reinforcement Learning
Shi, L., Li, S., Zheng, Q., Yao, M., & Pan, G. (2020) · IEEE Access · doi:10.1109/access.2020.3008735 · Cited by 27
“Cully et al. [9] propose a trial and error framework that allows robots to adapt to damage with high efficiency.”
BackgroundMethodFault-tolerant gait design for quadruped robots with one locked leg using the GF set theory
Chen, Z., Xi, Q., Gao, F., & Zhao, Y. (2022) · Mechanism and Machine Theory · doi:10.1016/j.mechmachtheory.2022.105069 · Cited by 23
Jegorova, M., Doncieux, S., & Hospedales, T. M. (2019) · doi:10.1109/icdl-epirob44920.2019 · Cited by 23
“Such behavioral repertoire-based approaches are emerging as promising techniques for robustly solving tasks [3], [4], [5].”
BackgroundInfluentialResilient machines through adaptive morphology
Häuser, H. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0076-6 · Cited by 24
“Previous work suggested purely algorithmic approaches to deal with damage by providing an alternative control strategy for the impaired body (2,3).”
BOP-Elites, a Bayesian Optimisation algorithm for Quality-Diversity searchPreprint
Kent, P. R. C., & Branke, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.04320 · Cited by 23
“In MAP-Elites, an evolutionary computation-based approach to QD [1], behavioural classifications are pre-defined instead of being emergent throughout the search process.”
BackgroundLearning contact-rich whole-body manipulation with example-guided reinforcement learning
Barreiros, J. A. L., Önol, A. Ö., Zhang, M., Creasey, S., Goncalves, A., Beaulieu, A., Bhat, A. D., Tsui, K., & Alspach, A. N. (2025) · Science Robotics · doi:10.1126/scirobotics.ads6790 · Cited by 46
Duarte, M., Gomes, J., Oliveira, S. M., & Christensen, A. L. (2016) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/2908812.2908855 · Cited by 22
“QD approaches have shown promising results in the evolution of a variety of locomotion behaviors for complex robots, such as hexapods [6, 4] and soft-robots [20].”
BackgroundMethodDynamics-Aware Quality-Diversity for Efficient Learning of Skill Repertoires
Lim, B., Grillotti, L., Bernasconi, L., & Cully, A. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9811559 · Cited by 28
“We define a primitive policy as a parametric low-level controller, as used in [4], [5].”
BackgroundMethodInfluentialSelf-citationFontaine, M. C., & Nikolaidis, S. (2022) · ACM Transactions on Human-Robot Interaction · doi:10.1145/3476412 · Cited by 28
“MAP-Elites [15, 69] is a popular QD algorithm that searches along a set of explicitly deined measures called behavior characteristics (BCs), which induce a Cartesian space called a behavior space.”
BackgroundMethodEfficient Pipeline Planning for Expedited Distributed DNN Training
Luo, Z., Yi, X., Long, G., Fan, S., Wu, C., Yang, J., & Lin, W. (2022) · IEEE Conference on Computer Communications (INFOCOM) · doi:10.1109/infocom48880.2022.9796787 · Cited by 26
“Deep learning has advanced various applications in a wide range of domains [1][2][3].”
BackgroundThe Emergence of Canalization and Evolvability in an Open-Ended, Interactive Evolutionary System
Huizinga, J., Stanley, K. O., & Clune, J. (2018) · Artificial Life · doi:10.1162/artl_a_00263 · Cited by 28
Self-citationHassaninia, I., Bostanabad, R., Chen, W., & Mohseni, H. (2017) · Scientific Reports · doi:10.1038/s41598-017-15601-4 · Cited by 22
Photomorphogenesis for robot self-assembly: adaptivity, collective decision-making, and self-repair
Soorati, M. D., Heinrich, M. K., Ghofrani, J., Zahadat, P., & Hamann, H. (2019) · Bioinspiration & Biomimetics · doi:10.1088/1748-3190/ab2958 · Cited by 24
Neurorobots as a Means Toward Neuroethology and Explainable AI
Chen, K., Hwu, T. J., Kashyap, H. J., Krichmar, J. L., Stewart, K. M., Xing, J., & Zou, X. (2020) · Frontiers in Neurorobotics · doi:10.3389/fnbot.2020.570308 · Cited by 24
“In locomotion experiments with a hexapod robot (see Figure 10G ), a neural network with prior knowledge of potential behaviors and their values was used to rapidly adapt the robot’s gait in response to a broken or altered leg (Cully et al., 2015).”
MethodMorphological Development in Robotic Learning: A Survey
Naya-Varela, M., Faíña, A., & Duro, R. J. (2021) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2021.3052548 · Cited by 28
Noisy-Input Entropy Search for Efficient Robust Bayesian OptimizationPreprint
Fröhlich, L. P., Klenske, E. D., Vinogradska, J., Daniel, C., & Zeilinger, M. N. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.02820 · Cited by 46
Hierarchically Organized Latent Modules for Exploratory Search in Morphogenetic SystemsPreprint
Etcheverry, M., Moulin-Frier, C., & Oudeyer, P. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.01195 · Cited by 38
“So far, this behavior space has either been hand-defined in robotic scenarios [16, 21] or unsupervisedly learned with deep auto-encoders directly from raw observations [15, 38, 50, 54, 55].”
BackgroundUntethered Microgrippers for Precision Medicine
Zhou, H., Zhang, S., Liu, Z., Chi, B., Li, J., & Wang, Y. (2023) · Small · doi:10.1002/smll.202305805 · Cited by 25
Self-adaptive piezoelectric vibration absorber with semi-passive tunable resonant shunts
Paixão, J., Foltête, E., Sadoulet-Reboul, É., Chevallier, G., & Cogan, S. (2024) · Journal of Sound and Vibration · doi:10.1016/j.jsv.2024.118424 · Cited by 21
Heterogeneous Multiagent Zero-Shot Coordination by Coevolution
Xue, K., Wang, Y., Guan, C., Yuan, L., Fu, H., Fu, Q., Qian, C., & Yu, Y. (2024) · IEEE Transactions on Evolutionary Computation · doi:10.1109/tevc.2024.3485177 · Cited by 27
“For example, diversity has been found helpful in fast adaptation [8], [42], robust training [59], [61], sparse or deceptive reward [7], [22], open-ended learning [47], [54], unsupervised environment designing [12], [36], and ZSC [31], [48], [62].”
Energy-Efficient Slithering Gait Exploration for a Snake-Like Robot Based on Reinforcement Learning
Bing, Z., Lemke, C., Jiang, Z., Huang, K., & Knoll, A. (2019) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2019/785 · Cited by 21
Designing Air Flow with Surrogate-Assisted Phenotypic Niching
Hagg, A., Wilde, D., Asteroth, A., & Bäck, T. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58112-1_10 · Cited by 28
Discovering evolutionary stepping stones through behavior domination
Meyerson, E., & Miikkulainen, R. P. (2017) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3071178.3071315 · Cited by 21
“[6, 25].”
A comparison of illumination algorithms in unbounded spaces
Vassiliades, V., Chatzilygeroudis, K. I., & Mouret, J.-B. (2017) · Genetic and Evolutionary Computation Conference Companion (GECCO Companion) · doi:10.1145/3067695.3082531 · Cited by 21
“e Multi-dimensional Archive of Phenotypic Elites (MAP-Elites) [1, 9] algorithm proposed a conceptually simpler approach: it discretizes the behavior space into a grid of k cells, storing in each cell the elite solution over the evolutionary generations.”
MethodSelf-citationLearning highly diverse robot throwing movements through quality diversity search
Kim, S., & Doncieux, S. (2017) · Genetic and Evolutionary Computation Conference Companion (GECCO Companion) · doi:10.1145/3067695.3082463 · Cited by 21
“An alternative approach consists in building a repertoire of behaviors in which to look for behaviors adapted to the context [1, 2].”
Generating and Adapting to Diverse Ad Hoc Partners in Hanabi
Canaan, R., Gao, X., Togelius, J., Nealen, A., & Menzel, S. (2022) · IEEE Transactions on Games · doi:10.1109/tg.2022.3169168 · Cited by 33
Human-Timescale Adaptation in an Open-Ended Task Space
Adaptive Agent Team, Bauer, J., Baumli, K., Baveja, S., Behbahani, F., Bhoopchand, A., Bradley-Schmieg, N., Chang, M., Clay, N., Collister, A., et al. (2023) · arXiv · doi:10.48550/arxiv.2301.07608 · Cited by 19
“Adaptation also plays a critical role in robotics, with agents trained to adapt to varying terrain (Clavera et al., 2019; Kumar et al., 2021) or damaged joints (Cully et al., 2015).”
BackgroundHuizinga, J., & Clune, J. (2022) · Evolutionary Computation · doi:10.1162/evco_a_00301 · Cited by 20
“By discretizing the behavior space, Cully et al. (2015) were able to evolve many different modes of behavior for a hexapod robot, which formed the basis of an intelligent trial-and-error algorithm that enabled the robot to quickly respond to damage.”
BackgroundSelf-citationResilient Robot Teams: a Review Integrating Decentralised Control, Change-Detection, and Learning
Bossens, D. M., Ramchurn, S. D., & Tarapore, D. (2022) · Current Robotics Reports · doi:10.1007/s43154-022-00079-4 · Cited by 27
Self-citationFactors Impacting Diversity and Effectiveness of Evolved Modular Robots
Pigozzi, F., Medvet, E., Bartoli, A., & Rochelli, M. (2023) · ACM Transactions on Evolutionary Learning and Optimization · doi:10.1145/3587101 · Cited by 25
“Finally, we experiment with an established quality-diversity algorithm, Multidimensional Archive of Phenotypic Elites (MAP-Elites, hence further abbreviated as ME) [Cully et al. 2015].”
Online Damage Recovery for Physical Robots with Hierarchical Quality-Diversity
Allard, M., Smith, S. C., Chatzilygeroudis, K. I., Lim, B., & Cully, A. (2023) · ACM Transactions on Evolutionary Learning and Optimization · doi:10.1145/3596912 · Cited by 26
“Robots can learn skills to solve various problems through methods coming from the field of robot learning [Akkaya et al. 2019; Cully et al. 2015; Miki et al. 2022].”
InfluentialSelf-citationWang, S., Ding, B., & Jin, Y. (2023) · IEEE Computational Intelligence Magazine · doi:10.1109/mci.2023.3277770 · Cited by 20
Automated Test Suite Generation for Software Product Lines Based on Quality-Diversity Optimization
Xiang, Y., Huang, H. P., Li, S., Li, M., Luo, C. W., & Yang, X. (2023) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3628158 · Cited by 22
Generating gaits for simultaneous locomotion and manipulation
Whitman, J. S., Su, S., Coros, S., Ansari, A. R., & Choset, H. (2017) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2017.8206099 · Cited by 20
“[11] generated a low dimensional map of behaviors which could then be used to quickly regenerate gaits online after the robot was damaged, but required two weeks of precomputation for each robot design.”
BackgroundEvolving a Repertoire of Controllers for a Multi-function Swarm
Engebråten, S. A., Moen, J., Yakimenko, O. A., & Glette, K. (2018) · Lecture notes in computer science · doi:10.1007/978-3-319-77538-8_49 · Cited by 20
“For this reason, we chose to explore the option of evolving controllers using the MAP-elites algorithm [8, 20].”
BackgroundMethodInteroceptive robustness through environment-mediated morphological development
Kriegman, S., Cheney, N., Corucci, F., & Bongard, J. (2018) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3205455.3205529 · Cited by 21
“Although generally absent from engineered systems (but see [3, 11]), the canonical form of robustness is seen to some extent in all organisms, and it comes from the act of development itself.”
BackgroundA Bioinspired Gait Transition Model for a Hexapod Robot
Chang, Q., & Mei, F. (2018) · Journal of Robotics · doi:10.1155/2018/2913636 · Cited by 20
Exploring Genetic Programming Systems with MAP-Elites
Dolson, E. L., Lalejini, A., & Ofria, C. (2019) · Genetic and evolutionary computation · doi:10.1007/978-3-030-04735-1_1 · Cited by 21
“Thus far, MAP-Elites had been used to evolve robot arms (Cully and Demiris, 2018), robot gaits (Cully et al., 2015), soft-bodied robots (Mouret and Clune, 2015), and neural networks for a computer vision task (Mouret and Clune, 2015).”
BackgroundMethodPrasad, V. K., & Bhavsar, M. D. (2020) · International Journal of E-Health and Medical Communications · doi:10.4018/ijehmc.2020070104 · Cited by 20
Hayakawa, T., Kamimura, T., Kaji, S., & Matsuno, F. (2020) · IEEE Transactions on Robotics · doi:10.1109/tro.2020.2992983 · Cited by 20
“According to the leg motors’ DOF configuration, legged robots can be classified into two groups: a mammal type, where the leg exists under the body like animals [13], and a sprawling type, where the leg exists on the side of the body like insects [12], [20], [22], [24], [25].”
BackgroundPedersen, J. W., & Risi, S. (2021) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3449639.3459317 · Cited by 32
“Several previous studies have focused on the ability of a robot show robustness in the face of a severe leg injury [9, 11, 35].”
Combining MAP-Elites and Incremental Evolution to Generate Gaits for a Mammalian Quadruped Robot
Nordmoen, J., Ellefsen, K. O., & Glette, K. (2018) · Lecture notes in computer science · doi:10.1007/978-3-319-77538-8_48 · Cited by 20
“QD algorithms like MAP-Elites [8] separates the behavior, of the system in question, from the fitness.”
BackgroundMethodEvolution of a functionally diverse swarm via a novel decentralised quality-diversity algorithm
Hart, E., Steyven, A., & Paechter, B. (2018) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3205455.3205481 · Cited by 24
“EDEA-fps Figure 9: Precision of a) swarm-maps at end generation b) active genomes in last generation (over 30 runs) on in future, in the face of varying environmental conditions (e.g. as described by [5] for a single robot). Figure 8 shows the swarmmap created from a single run of each of the EDQD algorithms at genera…”
BackgroundMethodInfluentialTang, Z., Tian, L., Xin, W., Wang, Q., Rus, D. L., & Laschi, C. (2026) · Science Advances · doi:10.1126/sciadv.aea3712 · Cited by 28
“In addition, the number of failed actuators reaches up to three of six actuators (one-half), which is a greater challenge than that in previous studies, where a six-legged robot was missing one leg (one-sixth) ( 46 ) and a hexapod robot was missing two legs (one-third) ( 47 ).”
Meta Reinforcement Learning of Locomotion Policy for Quadruped Robots With Motor Stuck
Chen, C., Li, C., Lu, H., Wang, Y., & Xiong, R. (2024) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2024.3424328 · Cited by 22
“Hence, it is crucial to develop an algorithm that can equip quadruped robots with fault tolerance [3], [4], [5], [6].”
Enhancing Divergent Search through Extinction Events
Lehman, J., & Miikkulainen, R. P. (2015) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/2739480.2754668 · Cited by 28
Rapid Phenotypic Landscape Exploration Through Hierarchical Spatial Partitioning
Smith, D., Tokarchuk, L. N., & Wiggins, G. A. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-45823-6_85 · Cited by 19
“MAP-Elites has been shown to be extendible to the real world application of robot controllers [2].”
BackgroundMethodLi, Y., Pu, Q., Li, S., Zhang, H., Wang, X., Yao, H., & Zhao, L. (2018) · Pattern Recognition Letters · doi:10.1016/j.patrec.2018.11.008 · Cited by 19
WatchPlant: Networked Bio-hybrid Systems for Pollution Monitoring of Urban Areas
Hamann, H., Bogdan, S., Díaz‐Espejo, A., García‐Carmona, L., Hernández‐Santana, V., Kernbach, S., Kernbach, A., Quijano-López, A., Salamat, B., & Wahby, M. (2021) · Conference on Artificial Life (ALIFE) · doi:10.1162/isal_a_00377 · Cited by 19
“For offline evolution we may start from a simple genetic algorithm but may also test state-of-the-art approaches, such as MAP-Elites (Mouret and Clune, 2015; Cully et al., 2015), especially when facing tradeoffs.”
MethodRobot End-Effector Mounted Camera Pose Optimization in Object Detection-Based Tasks
Roveda, L., Maroni, M., Mazzuchelli, L., Praolini, L., Shahid, A. A., Bucca, G., & Piga, D. (2021) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-021-01558-0 · Cited by 20
“Cully et al. [36] describes a trial-and-error algorithm that allows robots to adapt their behaviour in the presence of a damage; [37] employs constrained BO to select three force-controller parameters for combined position/interaction tasks; [38] proposes a methodology to achieve automatic tuning…”
MethodSchilling, M., & Melnik, A. (2018) · Advances in intelligent systems and computing · doi:10.1007/978-3-319-99316-4_36 · Cited by 19
Global Search with Bernoulli Alternation Kernel for Task-oriented Grasping Informed by SimulationPreprint
Antonova, R., Kokić, M., Stork, J. A., & Kragić, D. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1810.04438 · Cited by 19
“Previous work explored using kernels informed by simulation [2, 3] and by the domain dynamics [4], in some cases allowing to learn kernel transforms automatically [5].”
MethodModel-Based Quality-Diversity Search for Efficient Robot Learning
Keller, L., Tanneberg, D., Stark, S., & Peters, J. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9340794 · Cited by 28
Naderi, H., Shojaei, A., & Huang, L. (2026) · Eng—Advances in Engineering · doi:10.3390/eng7070358 · Cited by 18
Efficient reinforcement learning for humanoid whole-body control
Lober, R., Padois, V., & Sigaud, O. (2016) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2016.7803348 · Cited by 18
“Secondly, in Section III-C, we show that by choosing a more intuitive and low-dimensional task parameterization, we are able to use Bayesian optimization (BO), which has proven to be the technique of choice for optimizing expensive cost functions in few trials [10], as is the case with robotic experimentation [11], [1…”
MethodThe BesMan Learning Platform for Automated Robot Skill Learning
Gutzeit, L., Fabisch, A., Otto, M., Metzen, J. H., Hansen, J. M., Kirchner, F., & Kirchner, E. A. (2018) · Frontiers in Robotics and AI · doi:10.3389/frobt.2018.00043 · Cited by 18
“, 2013) and (Cully et al., 2015) have applied behavioral descriptions of the controller to be optimized to build and update an internal model guiding the search towards controllers that work well in reality without adapting the original simulation model.”
BackgroundLearning behaviour-performance maps with meta-evolution
Bossens, D. M., Mouret, J.-B., & Tarapore, D. (2020) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3377930.3390181 · Cited by 18
“Quality-diversity algorithms have been applied widely, with applications in artificial life [17], video game design [7, 14], automated image generation [20], design of robot morphologies and controllers [11, 19, 21], and behaviour adaptation [3, 13].”
MethodSelf-citationFast Game Content Adaptation Through Bayesian-based Player Modelling
González-Duque, M., Palm, R. B., & Risi, S. (2021) · IEEE Conference on Games (CoG) · doi:10.1109/cog52621.2021.9619018 · Cited by 25
Evolutionary online behaviour learning and adaptation in real robots
Silva, F., Correia, L. M. P., & Christensen, A. L. (2017) · Royal Society Open Science · doi:10.1098/rsos.160938 · Cited by 19
“It should be noted that our concept of behaviour-fitness map differs from the concept of behaviour performance map presented in [22], which is used to record the highest-performing individual found during evolution for each point in a discretized version of the behaviour space.”
A Survey of Behavior Learning Applications in Robotics -- State of the Art and PerspectivesPreprint
Fabisch, A., Petzoldt, C., Otto, M., & Kirchner, F. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.01868 · Cited by 24
“A kind of learned behavior that does not fit into any category because it is more general and can be used in combination with any underlying behavior is presented by Cully et al. (2015).”
BackgroundInfluentialDREAM Architecture: a Developmental Approach to Open-Ended Learning in RoboticsPreprint
Doncieux, S., Bredèche, N., Goff, L. K. L., Girard, B., Coninx, A., Sigaud, O., Khamassi, M., Díaz-Rodríguez, N., Filliat, D., Hospedales, T. M., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.06223 · Cited by 20
“A more general purpose and open-ended alternative is to exploit a behavioural repertoire [80] and extract from it an adequate policy [16].”
Proximal Policy Gradient Arborescence for Quality Diversity Reinforcement Learning
Batra, S., Tjanaka, B., Fontaine, M. C., Petrenko, A., Nikolaidis, S., & Sukhatme, G. S. (2023) · International Conference on Learning Representations · doi:10.48550/arxiv.2305.13795 · Cited by 20
“QD algorithms originated with the NSLC [28, 29] and MAP-Elites [35, 8] algorithms.”
MethodTraining Diverse High-Dimensional Controllers by Scaling Covariance Matrix Adaptation MAP-Annealing
Tjanaka, B., Fontaine, M. C., Lee, D. H., Kalkar, A., & Nikolaidis, S. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3313012 · Cited by 17
“For example, one popular approach in robotic locomotion has been to train a collection of neural network controllers to enable a walking robot to adapt to damage [1], [2], [3], [4].”
BackgroundA Soft Actor-Critic Approach for a Blind Walking Hexapod Robot with Obstacle Avoidance
Wang, L., Li, R., Huangfu, Z., Feng, Y., & Chen, Y. (2023) · Actuators · doi:10.3390/act12100393 · Cited by 17
“For the current hexapod robots, plenty of problems still exist [34].”
BackgroundQuantum robotics: a review of emerging trends
Yan, F., Iliyasu, A. M., Li, N., Salama, A., & Hirota, K. (2024) · Quantum Machine Intelligence · doi:10.1007/s42484-024-00225-5 · Cited by 22
“Already, robotics has been transformational in many industries, notably manufacturing as well as other wide-ranging benefits to society, such as in search and rescue, healthcare, and transportation to mention just a few (Cully et al. 2015).”
Cazenille, L., Chemtob, Y., Bonnet, F., Gribovskiy, A., Francesco Mondada, C., Bredèche, N., & Halloy, J. (2018) · Lecture notes in computer science · doi:10.1007/978-3-319-95972-6_9 · Cited by 17
“Additionally, our methodology could make use of global optimisation techniques designed to minimise the number of evaluations before reaching convergence, like Bayesian Optimisation [37], [13].”
The perception of a robot partner’s effort elicits a sense of commitment to human-robot interaction
Székely, M., Powell, H., Vannucci, F., Rea, F., Sciutti, A., & Michael, J. (2019) · Interaction Studies Social Behaviour and Communication in Biological and Artificial Systems · doi:10.1075/is.18001.sze · Cited by 17
“Indeed, this latter issue may be all the more acute for robots designed to engage in autonomous trial-and-error learning (Cully et al., 2015).”
BackgroundMeta-Active Learning in Probabilistically Safe Optimization
Schrum, M., Connolly, M. J., Cole, E. R., Ghetiya, M., Gross, R. E., & Gombolay, M. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3193497 · Cited by 20
Marsono, M., Yoto, Y., Suyetno, A., & Nurmalasari, R. (2021) · Journal of Robotics and Control (JRC) · doi:10.18196/26134 · Cited by 17
Hierarchical quality-diversity for online damage recovery
Allard, M., Smith, S. C., Chatzilygeroudis, K. I., & Cully, A. (2022) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3512290.3528751 · Cited by 21
“introduced the Intelligent Trial and Error (ITE) algorithm as an adaptive control method based on MAP-Elites and Gaussian processes for planning and adaptation [10].”
BackgroundMethodInfluentialSelf-citationGao, Y., Deng, H., Zhang, J., Shu, Q., Xu, Z., Cao, X., Wang, B., & Gong, X. (2022) · Advanced Materials Technologies · doi:10.1002/admt.202200047 · Cited by 17
ReaCog, a Minimal Cognitive Controller Based on Recruitment of Reactive Systems
Schilling, M., & Cruse, H. (2017) · Frontiers in Neurorobotics · doi:10.3389/fnbot.2017.00003 · Cited by 21
“Instead, central to the approach by Cully et al. (2015) is the idea of a behavioral parametrization which allows to characterize the currently experienced situation in a continuous, low dimensional space.”
BackgroundMethodEnhanced Optimization with Composite Objectives and Novelty Selection
Shahrzad, H., Fink, D., & Miikkulainen, R. (2018) · Conference on Artificial Life (ALIFE) · doi:10.1162/isal_a_00113 · Cited by 21
“These approaches have helped scale novelty search to more complex environments, including an array of control [2, 7, 29] and content generation [19, 21, 22] domains.”
BackgroundData-efficient neuroevolution with kernel-based surrogate models
Gaier, A., Asteroth, A., & Mouret, J.-B. (2018) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3205455.3205510 · Cited by 20
“teraction that is possible with the physical system. A common approach to optimization in computationally expensive domains is to use approximate models of the objective function, or surrogate models [2, 8, 10, 12]. These models are created through arXiv:1804.05364v2 [cs.NE] 17 Apr 2018 GECCO ’18, July 15–19, 2018, Ky…”
BackgroundMethodSelf-citationLearning Robust Task Priorities of QP-Based Whole-Body Torque-Controllers
Charbonneau, M., Modugno, V., Nori, F., Oriolo, G., Pucci, D., & Ivaldi, S. (2018) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2018.8624995 · Cited by 17
“In [10], prior knowledge from simulations was exploited to find acceptable behaviors on the real robot, in few trials.”
MethodSilva, J. F., Correia, L., & Christensen, A. (2016) · Scholarpedia · doi:10.4249/scholarpedia.33333 · Cited by 17
Training tactile sensors to learn force sensing from each other
Chen, Z., Ou, N., Zhang, X., Wu, Z., Zhao, Y., Wang, Y., Spyrakos-Papastavridis, E., Lepora, N. F., Jamone, L., Deng, J., et al. (2026) · Nature Communications · doi:10.1038/s41467-026-68753-1 · Cited by 24
“Robotscapableof perceivingandinteractingwiththeenvironmentare revolutionizing the healthcare, service, and manufacturing industries 1,2 .”
Modeling the behavior of mobile robots using genetic algorithms
Zarevich, A., Владимирович, М. Ф., Yagodkin, A., & Zolnikov, K. (2022) · Modeling of systems and processes · doi:10.12737/2219-0767-2022-15-3-7-16 · Cited by 16
A Calibration and Real-Time Object Matching Method for Heterogeneous Multi-Camera System
Li, C., & Cai, C. (2023) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2023.3260263 · Cited by 21
“I N RECENT years, with the rapid development of the semiconductor industry and the continuous improvement of computer performance, intelligent robots are gradually applied to various fields [1].”
BackgroundEvolutionary Action Selection for Gradient-Based Policy Learning
Ma, Y., Liu, T., Wei, B., Liu, Y., Xu, K., & Li, W. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-30111-7_49 · Cited by 21
Fault-tolerant gait design for quadruped robots with two locked legs using the GF set theory
Chen, Z., Xi, Q., Qi, C., Chen, X., Gao, Y., & Gao, F. (2024) · Mechanism and Machine Theory · doi:10.1016/j.mechmachtheory.2024.105592 · Cited by 16
Synergizing Quality-Diversity with Descriptor-Conditioned Reinforcement Learning
Faldor, M., Chalumeau, F., Flageat, M., & Cully, A. (2025) · ACM Transactions on Evolutionary Learning and Optimization · doi:10.1145/3696426 · Cited by 21
Self-citationYang, Y., Zhang, C., Liu, Y., Ning, J., & Guo, Y. J. (2024) · Swarm and Evolutionary Computation · doi:10.1016/j.swevo.2024.101732 · Cited by 16
Quality Diversity under Sparse Interaction and Sparse Reward: Application to Grasping in Robotics
Huber, J., Hélénon, F., Coninx, A., Amar, F. B., & Doncieux, S. (2026) · Evolutionary Computation · doi:10.1162/evco_a_00363 · Cited by 18
A Turing test for collective motion
Herbert‐Read, J. E., Romenskyy, M., & Sumpter, D. J. T. (2015) · Biology Letters · doi:10.1098/rsbl.2015.0674 · Cited by 16
Searching for Quality Diversity When Diversity is Unaligned with Quality
Pugh, J. K., Soros, L. B., & Stanley, K. O. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-45823-6_82 · Cited by 16
“The hope is that the results from this study will also provide guidance for other QD algorithms in the future, such as MAP-Elites [1, 12].”
BackgroundInfluentialA Brief Survey on the Role of Dimensionality Reduction in Manipulation Learning and Control
Ficuciello, F., Falco, P., & Calinon, S. (2018) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2018.2818933 · Cited by 18
“In [99], Gaussian processes are used to model the cost function instead of modeling the dynamical”
MethodDynamic mutation in MAP-Elites for robotic repertoire generation
Nordmoen, J., Samuelsen, E., Ellefsen, K. O., & Glette, K. (2018) · Conference on Artificial Life (ALIFE) · doi:10.1162/isal_a_00110 · Cited by 21
“Multidimensional Archive of Phenotypic Elites (MAP-Elites) is one such QD algorithm that has been used to great effect (Cully et al., 2015).”
MethodSparse reward exploration via novelty search and emitters
Paolo, G., Coninx, A., Doncieux, S., & Laflaquière, A. (2021) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3449639.3459314 · Cited by 24
“Since, many divergent search algorithms have been developed, using different mechanisms to drive the search: curiosity [34], empowerment [5], surprise [20], diversity [10, 11, 16, 32], and novelty [25].”
Few-Shot Quality-Diversity Optimization
Salehi, A., Coninx, A., & Doncieux, S. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3148438 · Cited by 20
“This effect seems to be much more taskname QD R [ 1] QD G [ 2] FAERY R [ 3] FAERY G [ 4] FAERY it [ 5] c µ [ 6] λ [ 7] M 1.0 3.9 1.0 0.0 80 100 100 10 10 1 QD R :percentageofsolvedenvironments(normalizedin [0 , 1] )whenusingQDoptimizationfromscratch.”
BackgroundMethodInfluentialDesign and complexity evaluation of a self-cleaning heat exchanger
Brooks, S., & Roy, R. (2022) · International Journal of Heat and Mass Transfer · doi:10.1016/j.ijheatmasstransfer.2022.122725 · Cited by 16
Morel, A., Kunimoto, Y., Coninx, A., & Doncieux, S. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9811837 · Cited by 20
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.