Crossing the Reality Gap: A Survey on Sim-to-Real Transferability of Robot Controllers in Reinforcement Learning
Open accessThe DOI matches its Crossref record; 133 of the 173 references it cites were found in a registry or an index; 30 of the 71 sources its claims cite were checked for support.
Abstract
The growing demand for robots able to act autonomously in complex scenarios has widely accelerated the introduction of Reinforcement Learning (RL) in robots control applications. However, the trial and error intrinsic nature of RL may result in long training time on real robots and, moreover, it may lead to dangerous outcomes. While simulators are useful tools to accelerate RL training and to ensure safety, they often are provided only with an approximated model of robot dynamics and of its interaction with the surrounding environment, thus resulting in what is called the reality gap (RG): a mismatch of simulated and real control-law performances caused by the inaccurate representation of the real environment in simulation. The most undesirable result occurs when the controller learnt in simulation fails the task on the real robot, thus resulting in an unsuccessful sim-to-real transfer. The goal of the present survey is threefold: (1) to identify the main approaches to face the RG problem in the context of robot control with RL, (2) to point out their shortcomings, and (3) to outline new potential research areas.
Overview
CoChat Summary
TL;DR
This survey reviews sim-to-real transferability of RL robot controllers and categorizes reality-gap approaches into domain randomisation, adversarial RL, and transfer learning. It finds no task-independent solution or standard reality-gap metric, and notes many methods lack real-robot validation.
Key findings
- The survey identifies three main categories for addressing the reality gap: domain randomisation (DR), adversarial RL (ARL), and transfer learning (TL). Across the surveyed articles, 11 use DR, 2 use ARL, and 7 use TL (Tables 2–5).
- DR and ARL treat sim-to-real transferability as robustness to model perturbations, but the paper argues that robustness to chosen disturbances does not guarantee transferability if those disturbances do not correctly represent simulator inaccuracies.
- TL approaches appear promising because they involve interaction with the real system during training, but they often lack robustness assessment and may be inefficient due to two successive or simultaneous training phases.
- Open challenges include the absence of a general task-independent approach, the need for an index to reveal and quantify the reality gap, and the sample efficiency of RL-based methods.
Methods
- Literature survey: included articles had to explicitly address the reality gap, deal with robot control applications, and employ RL techniques; the survey then categorizes them into DR, ARL, and TL.
- The paper provides a formal RL/control framework and summarizes each surveyed article by algorithm, task, simulator, and whether evaluation was sim-to-sim and/or sim-to-real.
- Qualitative synthesis; no primary experiments or quantitative meta-analysis are conducted.
Limitations
- Stated: task diversity and the lack of a common theoretical framework prevent systematic comparison; many proposed approaches were not experimentally evaluated in an actual sim-to-real scenario; each approach appears tailored to a specific task; a general task-independent approach is still missing; sample efficiency remains a significant open problem.
- Evident: the survey does not report a systematic search protocol, databases, or quantitative meta-analysis, so its categorization and counts depend on the selected articles.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
18 Valid · 11 Not valid · 38 Not checked · 44 Own results and interpretations
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“The analysed task is a Marble maze game driven by a robotic arm and the results show that there is a trade-off between controller robustness and ne-tuning steps.”
What the sources say
- Disputedcites [56] [56] J. Van Baar, A. Sullivan, R. Cordorel, D. Jha, D. Romeres, and D. Nikovski, ''Sim-to-real transfer learning using robusti ed controllers in robotic tasks involving complex dynamics,'' in Proc. Int. Conf. Robot. Automat. (ICRA) , May 2019, pp. 6001 6007. <!-- image --> IEEEAccess" - sourceExperimental — accuracy not yet measured
Section 5.4 (p. 8) describes a 3D maze-navigation game controlled through movement and looking actions, not a marble maze driven by a robotic arm. The cited paper discusses transfer learning and finetuning, but the passages do not report the claimed trade-off between controller robustness and fine-tuning steps. (LLM, conf=0.84)
Checked by a reasoning model — may be wrong.
plainstates an associationSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Disputed
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“However, the authors assume that, when the consequences of a control input applied in simulation differ from those applied on the real robot, real observations should match simulated ones, and that is not always true.”
What the sources say
- Disputedcites [49] [49] P. Christiano, Z. Shah, I. Mordatch, J. Schneider, T. Blackwell, J. Tobin, P. Abbeel, and W. Zaremba, ''Transfer from simulation to real world through learning deep inverse dynamics model,'' 2016, arXiv:1610.03518 . - sourceExperimental — accuracy not yet measured
The Experiments section (p. 4) says that visual discrepancies between simulated and real observations are obvious, rather than assuming they should match. The Introduction (p. 2) describes training a new network on the real robot after training in simulation. The paper does not make the claimed assumption about observations when control inputs have different consequences. (LLM, conf=0.83)
Checked by a reasoning model — may be wrong.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Disputed
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“On the contrary, their evaluations have shown that controllers learnt in simulation with large observation space lead to bad results when transferred to real robot.”
What the sources say
- Partly supportedcites [54] [54] J. Tan, T. Zhang, E. Coumans, A. Iscen, Y. Bai, D. Hafner, S. Bohez, and V. Vanhoucke, ''Sim-to-real: Learning agile locomotion for quadruped robots,'' in Proceedings of Robotics: Science and Systems . Pittsburgh, PA, USA: RSS Foundation, Jun. 2018, pp. 10 20. - sourceExperimental — accuracy not yet measured
The ablation study (4 Experiments, pp. 7–8) found that a baseline controller trained in simulation failed the full task on the real robot. But it attributes that failure to the absence of domain randomisation, not to a large observation space; the fully randomised controller completed all standard real-world trials. (LLM, conf=0.85)
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“In particular, when the cGAN during training is fed with images sampled only from non-successful trajectories, the nal controller results in an unsatisfactory transfer.”
What the sources say
- Could not verifycites [46] [46] S. James, P. Wohlhart, M. Kalakrishnan, D. Kalashnikov, A. Irpan, J. Ibarz, S. Levine, R. Hadsell, and K. Bousmalis, ''Sim-to-real via sim-tosim: Data-ef cient robotic grasping via randomized-to-canonical adaptation networks,'' in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. , Jun. 2019, pp. 12627 12637. - sourceExperimental — accuracy not yet measured
Section 4.1 (p. 4) says the cGAN’s training images were sampled from QT-Opt trajectories, but does not describe training it only on unsuccessful trajectories. Section 5.2 (p. 7) reports transfer performance for other training approaches, not the claimed unsuccessful-trajectory condition. Searches for failed or unsuccessful trajectory sampling found no result addressing its effect on transfer. (LLM, conf=0.75)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
Every source the paper cites for this claim was checked and supports it.
In the paper
“Besides, they show that heavy randomisation can lead to unsuccessful transfers.”
What the sources say
- Supportedcites [45] [45] J. Matas, S. James, and A. J. Davison, ''Sim-to-real reinforcement learning for deformable object manipulation,'' in Proc. 2nd Annu. Conf. Robot Learn. (CoRL) , Z rich, Switzerland, Oct. 2018, pp. 734 743. [Online]. Available: https://proceedings.mlr.press/v87/matas18a.html - source
Section 5.3 Sim-to-real transfer (p. 8) states that across levels of domain randomisation, 'heavy randomisation can be detrimental to learning', and that sampling texture colours uniformly over all colours made the agent's post-transfer performance significantly worse. This is exactly what the citing sentence attributes, including the transfer framing (worse performance after the sim-to-real transfer). (LLM, conf=0.90)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
33 more claims · 5 with their cited sources checked
Create a free account to see all 111 claims and to verify the 12 references still open.
- It may state two separate claims as one. 37
- Probably not a substantial claim. 57
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ V. CONCLUSION AND OPEN CHALLENGES · p. 14
“When performances of the controller in the simulator and in real robot are different, a RG exists and the controller is not sim-to-real transferable.”
§ III. REINFORCEMENT LEARNING IN ROBOTICS · p. 5
“Note that, in general, it is not required that the controller behaves identically when applied to E and E 0 , but, rather, that it is E 0 -toE transferable, i.e., sim-to-real transferable. In practice, this requirement translates to a (subjectively) properly bounded RG.”
Citations
Cites
Human-level control through deep reinforcement learning
Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A., Riedmiller, M., Fidjeland, A. K., Ostrovski, G., et al. (2015) · Nature · doi:10.1038/nature14236 · Cited by 30,357 · cited 6×
“We categorise the different TL approaches based on: • the kind of information passed through u , weights of an image processing net, state-control input pairs, policy parameters or even the policy itself); • the transfer timing of u ( k ) exch , either continuous (C-TL) or one shot (1-TL), the latter generally occurri…”
MethodInfluentialReinforcement Learning: An Introduction
Sutton, R. S., & Barto, A. G. (1998) · IEEE Transactions on Neural Networks · doi:10.1109/tnn.1998.712192 · Cited by 27,225 · cited 5×
“RL approaches can be categorised in three main categories [1]:”
BackgroundMethodInfluentialReinforcement Learning: An Introduction
(2005) · IEEE Transactions on Neural Networks · doi:10.1109/tnn.2004.842673 · Cited by 24,496
Image-to-Image Translation with Conditional Adversarial Networks
Isola, P., Zhu, J.-Y., Zhou, T., & Efros, A. A. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.632 · Cited by 22,930 · cited 1×
“The map is learnt by using an image-conditioned Generative Adversarial Network (cGAN) [98], able to map an image of a domain D into an adapted image of the canonical domain Dc; i.”
MethodLabaca-Castro, R. (2023) · Machine Learning under Malware Attack · doi:10.1007/978-3-658-40442-0_9 · Cited by 19,272
GAN(Generative Adversarial Nets)
(2017) · Journal of Japan Society for Fuzzy Theory and Intelligent Informatics · doi:10.3156/jsoft.29.5_177_2 · Cited by 19,270
Proximal Policy Optimization Algorithms
Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017) · arXiv · doi:10.48550/arxiv.1707.06347 · Cited by 10,625
Explaining and Harnessing Adversarial Examples
Goodfellow, I. J., Shlens, J., & Szegedy, C. (2014) · arXiv · doi:10.48550/arxiv.1412.6572 · Cited by 10,381 · cited 2×
“The agent uses a curriculum learning approach [91], in which ξ (cid:48) , υ (cid:48) ( k ) , and ζ (cid:48) ( k ) alternately assume the form of isometrically scaled versions of Fast Gradient Sign Method (FGSM) [92].”
MethodIntriguing properties of neural networks
Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I., & Fergus, R. (2013) · arXiv · doi:10.48550/arxiv.1312.6199 · Cited by 8,143 · cited 1×
“The choice of adversarial perturbations is motivated by the fact that by employing them, the resulting models are likely to generalise well [93].”
BackgroundAsynchronous Methods for Deep Reinforcement Learning
Mnih, V., Badia, A. P., Mirza, M., Graves, A., Lillicrap, T. P., Harley, T., Silver, D., & Kavukcuoglu, K. (2016) · arXiv · doi:10.48550/arxiv.1602.01783 · Cited by 7,132 · cited 1×
“[56] shows the benefits of using DR and Asynchronous Actor Critic Agent (A3C) algorithm [86] (on-policy) for learning the controller, with respect to not using DR.”
Method
Research works printed but not linked: 10
- tedrake, 2005, Learning to walk in 20 minutes, Proc 14th Yale Workshop Adapt Learn Syst, 95585, 1412
- bertsekas, 2019, REINFORCEMENT LEARNING AND OPTIMAL CONTROL
- coumans, 2016, Pybullet a python module for physics simulation for games robotics and machine learning
- coumans, 2016, Bullet Physics SDK
- 2020, Blender Open Source 3D Modeling Suit
- 2020, Simspark
- 2020, Gazebo-robot simulation made easy
- 2001, Vortex Studio Real-time simulation and visualization software for system-level modeling of mechatronics and mechanical equipment
- brockman, 2016, OpenAI gym, arXiv 1606 01540 [cs]
- 2020, Poppy Project Open Source Robotics Platform
Cited by
Experiment-free exoskeleton assistance via learning in simulation
Luo, S., Jiang, M., Zhang, S., Zhu, J., Yu, S., Dominguez Silva, I., Wang, T., Rouse, E., Zhou, B., Yuk, H., et al. (2024) · Nature · doi:10.1038/s41586-024-07382-4 · Cited by 166
Hu, X., Li, S., Huang, T., Tang, B., Huai, R., & Chen, L. (2023) · IEEE Transactions on Intelligent Vehicles · doi:10.1109/tiv.2023.3312777 · Cited by 240
“At the same time, it solves two problems of traditional simulators [88], [89]: 1) simulation testing is not equivalent to real end-to-end testing, and 2) factors such as weather, climate, and lighting are not fully covered.”
Backgrounddel Real, A. J., Andreiana, D. S., Roldán, Á. O., Bustos, A. H., & Acevedo, L. (2022) · Applied Sciences · doi:10.3390/app122312377 · Cited by 128
“…are aggregating demonstration data to the experience of the agent in a model-based RL algorithm to learn better models; including scripted policies to initialise the training, such as in QT-Opt [113] and reward shaping provides additional guidance to exploration, boosting the learning process.”
MethodLee, H., Jeon, J., Lee, D., Park, C., Kim, J., & Lee, D. (2023) · Automation in Construction · doi:10.1016/j.autcon.2023.105060 · Cited by 57
Sim-to-real transfer and reality gap modeling in model predictive control for autonomous driving
Daza, I. G., Izquierdo, R., Martínez, L. M., Benderius, O., & Llorca, D. F. (2022) · Applied Intelligence · doi:10.1007/s10489-022-04148-1 · Cited by 58
“[13] conducted a recent survey about closing the reality gap and reduce the sim-to-real transferability when developing a robot controller based on reinforcement learning.”
BackgroundChakraborty, N., Ornik, M., & Driggs-Campbell, K. (2025) · ACM Computing Surveys · doi:10.1145/3716846 · Cited by 95
“Some knowledge grounding approaches prompt LLMs to generate code to directly query information from databases.”
Spatiotemporal Attention Enhances Lidar-Based Robot Navigation in Dynamic Environments
de Heuvel, J., Zeng, X., Shi, W., Sethuraman, T. V., & Bennewitz, M. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3373988 · Cited by 58
“These approaches suffer from a reality gap that hinders generalization to the real world [9], [10].”
BackgroundYasutomi, A. Y., Ichiwara, H., Ito, H., Mori, H., & Ogata, T. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3243526 · Cited by 38
“Another strategy to reduce task completion time is to extensively train DRL algorithms in simulation environments and transfer the trained model to the real world [8].”
BackgroundRobotic Disassembly Task Training and Skill Transfer Using Reinforcement Learning
Qu, M., Wang, Y., & Pham, D. T. (2023) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2023.3242831 · Cited by 42
“Forexample,tobypassthecommonissuesofRL associatedwithlackofdataandunsafeexplorationbyrobots, simulatorshavebeenutilizedtogenerateskillsforvirtualrobots; the virtually generated skills can subsequently be transferred to the physical world [25].”
Arcieri, G., Hoelzl, C., Schwery, O., Straub, D., Papakonstantinou, K. G., & Chatzi, E. (2024) · Machine Learning · doi:10.1007/s10994-024-06559-2 · Cited by 38
Trajectory Planning With Deep Reinforcement Learning in High-Level Action Spaces
Williams, K., Schlossman, R., Whitten, D., Ingram, J., Musuvathy, S., Pagan, J. E. G., Williams, K. A., Green, S., Patel, A., Mazumdar, A., et al. (2022) · IEEE Transactions on Aerospace and Electronic Systems · doi:10.1109/taes.2022.3218496 · Cited by 34
“So-called sim2real techniques, such as domain randomization, adversarial RL, and transfer learning [42], have been developed to facilitate transfer to hardware when applicable.”
BackgroundSim-to-real transfer of active suspension control using deep reinforcement learning
Wiberg, V., Wallin, E., Fälldin, A., Semberg, T., Rossander, M., Wadbro, E., & Servin, M. (2024) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2024.104731 · Cited by 46
Birchler, C., Khatiri, S., Rani, P., Kehrer, T., & Panichella, S. (2025) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3711906 · Cited by 42
“With the boost of Evolutionary Robotics and the application of reinforcement learning in designing robotic con-trol systems in recent decades, practitioners rely more and more on simulations to evaluate their designs,i.e., test subjects [22, 40].”
Reinforcement learning applications in water resource management: a systematic literature review
Kåge, L., Milić, V., Andersson, M., & Wallén, M. (2025) · Frontiers in Water · doi:10.3389/frwa.2025.1537868 · Cited by 42
How Does Simulation-Based Testing for Self-Driving Cars Match Human Perception?
Birchler, C., Mohammed, T. K., Rani, P., et al. (2024) · Proceedings of the ACM on software engineering. · doi:10.1145/3643768 · Cited by 28
“To that end, in our research, we investigated when and why the safety metrics of simulation-based test cases of SDCs match human perception.”
Croll, H. C., Ikuma, K., Ong, S. K., & Sarkar, S. (2024) · Water Research · doi:10.1016/j.watres.2024.122179 · Cited by 38
Evolving Hebbian Learning Rules in Voxel-Based Soft Robots
Ferigo, A., Iacca, G., Medvet, E., & Pigozzi, F. (2022) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2022.3226556 · Cited by 35
“Visual inspection is fundamental since behaviors are a strong indicator of a possible reality gap [76], [77].”
BackgroundSelf-citationNaturally Interpretable Control Policies via Graph-Based Genetic Programming
Nadizar, G., Medvet, E., & Wilson, D. G. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-56957-9_5 · Cited by 30
Self-citationFusion of Computer Vision and AI in Collaborative Robotics: A Review and Future Prospects
Cohen, Y., Biton, A., & Shoval, S. (2025) · Applied Sciences · doi:10.3390/app15147905 · Cited by 36
Learning-Based Navigation and Collision Avoidance Through Reinforcement for UAVs
Azzam, R., Chehadeh, M. S., Hay, O. A., Humais, M. A., Boiko, I. M., & Zweiri, Y. (2023) · IEEE Transactions on Aerospace and Electronic Systems · doi:10.1109/taes.2023.3294889 · Cited by 25
Knowledge Transfer for Cross-Domain Reinforcement Learning: A Systematic Review
Serrano, S. A., Martínez-Carranza, J., & Sucar, L. E. (2024) · IEEE Access · doi:10.1109/access.2024.3435558 · Cited by 18
“Some ideas to tackle the sample complexity of real-world robots are to combine sim-to-real approaches [46] with a refinement/adaptation phase of policies pretrained in simulation or extensive data sets [141], [142], [143], [144], [145].”
Domain Randomization for Sim2real Transfer of Automatically Generated Grasping Datasets
Huber, J., Hélénon, F., Watrelot, H., Amar, F. B., & Doncieux, S. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610677 · Cited by 48
“The difference between simulation and reality ( reality gap ) is well-known as a major challenge in robotics [43].”
BackgroundIncreasing the Flexibility of Hydropower with Reinforcement Learning on a Digital Twin Platform
Tubeuf, C., Birkelbach, F., Maly, A., & Hofmann, R. (2023) · Energies · doi:10.3390/en16041796 · Cited by 31
Juříček, M., Parák, R., & Kůdela, J. (2023) · Computation · doi:10.3390/computation11120245 · Cited by 27
“The reality gap is a frequent phenomenon and is still one of the main obstacles to progress in utilizing not only evolutionary computation [151] but other techniques as well [152].”
Shakerimov, A., Alizadeh, T., & Varol, H. A. (2023) · IEEE Access · doi:10.1109/access.2023.3339568 · Cited by 39
“During training, a learning algorithm updates the control policy towards obtaining maximal cumulative reward under every encountered disturbance scenario [33].”
BackgroundMethodSmall-scale self-driving cars: A systematic literature review
Caleffi, F., Rodrigues, L. D. S., Stamboroski, J. D. S., & Pereira, B. M. (2024) · Journal of Traffic and Transportation Engineering (English Edition) · doi:10.1016/j.jtte.2023.09.005 · Cited by 31
Park, S.-Y., Lee, C., Kim, H., & Ahn, S. (2024) · IEEE Access · doi:10.1109/access.2024.3359268 · Cited by 21
“Several methods using simulation and optimization algorithms have been proposed to solve these problems in real robots [9], [10].”
BackgroundMethodFactors 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 hypothesize that the diversity induced by SE-b and similar EAs might be beneficial for fighting the reality gap problem, which affects the field of automatic design of robots, not only when employing ER techniques [Salvato et al. 2021].”
Self-citationTowards applied swarm robotics: current limitations and enablers
Kegeleirs, M., & Birattari, M. (2025) · Frontiers in Robotics and AI · doi:10.3389/frobt.2025.1607978 · Cited by 33
In silico simulation: a key enabling technology for next-generation intelligent surgical systems
Killeen, B. D., Cho, S. M., Armand, M., Taylor, R. H., & Unberath, M. (2023) · Progress in Biomedical Engineering · doi:10.1088/2516-1091/acd28b · Cited by 22
“This approach is already prominent in related problems of general robotic manipulation [41, 42] and self-driving [43], each of which contends with challenges specific to the given task, environment, and data modality.”
Nadizar, G., Medvet, E., Nichele, S., & Pontes-Filho, S. (2023) · Applied Soft Computing · doi:10.1016/j.asoc.2023.110610 · Cited by 16
Self-citationLiang, X., Liu, F., Zhang, Y., Li, Y., Lin, S., & Yip, M. C. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10610263 · Cited by 22
“on closing the gap through effective policy transfer in a reinforcement learning (RL) setup, as reviewed in [14], [15].”
BackgroundImitation learning of a model predictive controller for real-time humanoid robot walking
Porto, V. G. B. D. A., Melo, D. C., Máximo, M. R. O. A., & Afonso, R. J. M. (2025) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2024.109919 · Cited by 12
Learning Interactive Real-World Simulators
Yang, S., Du, Y., Ghasemipour, K., Tompson, J., Kaelbling, L., Schuurmans, D., & Abbeel, P. (2023) · arXiv · doi:10.48550/arxiv.2310.06114 · Cited by 346
“• We formulate the action-in-video-out framework as an observation prediction model conditioned on finite history and parametrized by a video diffusion model.”
MethodGP for Continuous Control: Teacher or Learner? The Case of Simulated Modular Soft Robots
Medvet, E., & Nadizar, G. (2024) · Genetic and evolutionary computation · doi:10.1007/978-981-99-8413-8_11 · Cited by 14
Self-citationSontakke, N., Chae, H., Lee, S. J., Huang, T., Hong, D. W., & Hal, S. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10342062 · Cited by 27
“However, successfully deploying these controllers on hardware is still an active area of research [26], [27], which is not straightforward due to the discrepancy between the simulation and the real world [6].”
BackgroundLearning Volt-VAR Droop Curves to Optimally Coordinate Photovoltaic (PV) Smart Inverters
Glover, D., & Dubey, A. (2024) · IEEE Transactions on Industry Applications · doi:10.1109/tia.2024.3472655 · Cited by 11
“Learned policies suffer from limited transferability and policy degradation when exposed to unseen data distributions online, known as the simulation-to-reality gap [25], [26].”
Applications of deep reinforcement learning in nuclear energy: A review
Liu, Y., Wang, B., Tan, S., Li, T., Lv, W., Niu, Z., Li, J., Gao, P., & Tian, R. (2024) · Nuclear Engineering and Design · doi:10.1016/j.nucengdes.2024.113655 · Cited by 12
Medvet, E., Nadizar, G., & Pigozzi, F. (2022) · Genetic and Evolutionary Computation Conference Companion (GECCO Companion) · doi:10.1145/3520304.3533967 · Cited by 13
“This problem is known as reality gap and affects many kinds of adaptation, including evolutionary optimization [8, 13] and reinforcement learning [34, 38].”
BackgroundSelf-citationA Fully-distributed Shape-aware Neural Controller for Modular Robots
Nadizar, G., Medvet, E., Walker, K., & Risi, S. (2023) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3583131.3590419 · Cited by 15
“However, it turns out that such richness often results, upon optimization, in vibrating behaviors [17] that would be poorly effective in reality—an instance of the so-called reality gap problem [20, 27, 30].”
BackgroundSelf-citationZhang, Y., Thor, M., Dilokthanakul, N., Dai, Z., & Manoonpong, P. (2023) · Neural Networks · doi:10.1016/j.neunet.2023.08.030 · Cited by 11
Kegeleirs, M., Ramos, D. G., Hasselmann, K., Garattoni, L., Francesca, G., & Birattari, M. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3360013 · Cited by 12
“Several ideas have been proposed to address this problem—for recent reviews focusing on reinforcement learning, see [1], [2].”
BackgroundInfluentialSim–Real Mapping of an Image-Based Robot Arm Controller Using Deep Reinforcement Learning
Sasaki, M., Muguro, J. K., Kitano, F., Njeri, W., & Matsushita, K. (2022) · Applied Sciences · doi:10.3390/app122010277 · Cited by 13
“As noted by [6], these approaches are inherently data‐intensive, thus limiting the practical applicability of such methods.”
BackgroundMethodThe 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
“Exposing evolving robots to variable conditions is also necessary to cross the reality gap, that is, to transfer robots trained in simulation into hardware successfully (Jakobi et al., 1995; Koos et al., 2012; Sadeghi and Levine, 2016; Salvato et al., 2021).”
BackgroundApproaching Collaborative Manipulation by Pushing-Grasping Fusion
Phan, L. A. D., Phan, D. Q., Bui, T. T., Le, P. H., Tran, D. T., Lee, J., & Ngo, H. Q. T. (2025) · IEEE Access · doi:10.1109/access.2025.3574018 · Cited by 13
“In general, there were a group of researchers whom tried to translate the original results of training procedure to the real-world applications without any modifications or least change [13].”
Multi-UAV Reinforcement Learning With Realistic Communication Models: Recent Advances and Challenges
Cattai, T., Frattolillo, F., Lacava, A., Raut, P., Simonjan, J., D’Oro, S., Melodia, T., Vinogradov, E. A., Natalizio, E., Colonnese, S., et al. (2025) · IEEE Open Journal of Vehicular Technology · doi:10.1109/ojvt.2025.3586774 · Cited by 23
“The most important is denoted as reality gap , which is the mismatch between the simulated and the real control-law caused by small modeling inaccuracies in the simulator [42].”
A Survey on UxV Swarms and the Role of Artificial Intelligence as a Technological Enabler
Dimos, A., Skoutas, D. N., Νομικός, Ν., & Skianis, C. (2025) · Drones · doi:10.3390/drones9100700 · Cited by 13
Al-Sharman, M., Edes, L., Sun, B., Jayakumar, V., Tahir, H., Daoud, M. A., Emran, B. J., Rayside, D., & Melek, W. (2025) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2025.3646982 · Cited by 28
“This survey reviews various aspects related to challenges associated with decision-making at unsignalized intersections with a focus on learning-based schemes.”
Reinforcement Learning in Few-Shot Scenarios: A Survey
Wang, Z., Fu, Q., Chen, J., Wang, Y., Lu, Y., & Wu, H. (2023) · Journal of Grid Computing · doi:10.1007/s10723-023-09663-0 · Cited by 13
“The term reality gap [22,23] is used to describe the negative results of learning algorithms applied from the simulated world to the real world.”
BackgroundMock, J. W., & Muknahallipatna, S. (2024) · Journal of Intelligent Learning Systems and Applications · doi:10.4236/jilsa.2024.162003 · Cited by 15
Fine-Tuning Multimodal Transformer Models for Generating Actions in Virtual and Real Environments
Staroverov, A., Gorodetsky, A. S., Krishtopik, A., Izmesteva, U. A., Yudin, D. A., Kovalev, A., & Panov, A. (2023) · IEEE Access · doi:10.1109/access.2023.3334791 · Cited by 11
“Thereare many methods for solving the problem of transferring the skills of models trained in simulators to control real robots (the so-called sim-to-real problem) [28], [29].”
Research on Robot Massage Force Control Based on Residual Reinforcement Learning
Xiao, M., Zhang, T., Zou, Y., Chen, S., & Wu, W. R. (2023) · IEEE Access · doi:10.1109/access.2023.3347416 · Cited by 9
“The reinforcement learning agent needs to improve the policy through trial and error; therefore, in the actual interaction process, multiple experiments are required to achieve the desired result [35].”
BackgroundHong, Y., Zhao, H., Li, X., Chen, Y., Xia, G., & Ding, H. (2025) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2025.3572722 · Cited by 8
“The related limitations are discussed in [44], [45], with existing methods including: building high-quality assets for improved simulation accuracy [46], using the policy action integrator to minimize errors [47], and dynamically adjusting compliance gains in reality [48], etc.”
Outline of an evolutionary morphology generator towards the modular design of a biohybrid catheter
Tsompanas, M., & Balaž, I. (2024) · Frontiers in Robotics and AI · doi:10.3389/frobt.2024.1337722 · Cited by 8
Transition Control of a Double-Inverted Pendulum System Using Sim2Real Reinforcement Learning
Lee, T., Ju, D., & Lee, Y. S. (2025) · Machines · doi:10.3390/machines13030186 · Cited by 14
FPGA-Accelerated Sim-to-Real Control Policy Learning for Robotic Arms
Guo, C., & Luk, W. W. (2024) · IEEE Transactions on Circuits & Systems II Express Briefs · doi:10.1109/tcsii.2024.3353690 · Cited by 8
“A viable alternative is sim-to-real robot learning [2], [3], where the reinforcement learning algorithm initially trains using virtual robots in simulated environments before applying the learned policy to physical robots.”
Czarnecki, T., Stawowy, M., & Kadłubowski, A. (2024) · Applied Sciences · doi:10.3390/app15010179 · Cited by 7
Albeaik, S., Wu, T., Vurimi, G., Chou, F., Lu, X., & Bayen, A. M. (2022) · Journal of Field Robotics · doi:10.1002/rob.22131 · Cited by 8
“However, for real ‐ world robotics, where data could be challenging to collect, and only approximations of systems dynamics and states can be captured, success has been limited (Liu et al., 2020; Salvato et al., 2021; Zhao et al., 2020).”
BackgroundLi, S., López‐Benítez, M., Lim, E. G., Ma, F., Cao, M., Yu, L., & Qin, X. (2025) · Drones · doi:10.3390/drones9110752 · Cited by 9
Curriculum Design and Sim2Real Transfer for Reinforcement Learning in Robotic Dual-Arm Assembly
Wrede, K., Zarnack, S., Lange, R., Donath, O., Wohlfahrt, T., & Feldmann, U. (2024) · Machines · doi:10.3390/machines12100682 · Cited by 8
Deep Reinforcement Learning-Assisted Teaching Strategy for Industrial Robot Manipulator
Šimon, J., Gogolák, L., & Sárosi, J. (2024) · Applied Sciences · doi:10.3390/app142310929 · Cited by 8
Combining Reinforcement Learning and Lazy Learning for Faster Few-Shot Transfer Learning
Lončarević, Z., Simonič, M., Ude, A., & Gams, A. (2022) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids53995.2022.10000095 · Cited by 8
“This difference is termed the reality gap (RG) [6].”
BackgroundX-62 VISTA Capabilities and Architecture
Cotting, M. C., Stephens, S. S., Cole, J., et al. (2023) · AIAA SciTech Forum · doi:10.2514/6.2023-1744 · Cited by 6
Sivashangaran, S., Khairnar, A., & Eskandarian, A. (2023) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2023.12.069 · Cited by 9
Centralized Coordination of DER Smart Inverters using Deep Reinforcement Learning
Glover, D., & Dubey, A. (2023) · IEEE Industry Applications Society Annual Meeting (IAS) · doi:10.1109/ias54024.2023.10406599 · Cited by 6
“However, they suffer from lack of exploration and safety guarantees, scalability issues and policy degradation over time leading to poor online transfer, known as the simulation-to-reality gap [13].”
BackgroundCho, D.-S., Cho, J.-M., & Kim, W.-T. (2024) · IEEE Transactions on Reliability · doi:10.1109/tr.2024.3434606 · Cited by 5
“If synchronization fails, it may cause unintended controls by the HLC-SW due to the sim2real gap [44].”
Safedrive dreamer: Navigating safety–critical scenarios in autonomous driving with world models
Li, H., Peng, T., Wang, B., Zhang, R., Gao, B., Qiao, N., Guan, Z., Li, J., & shi, T. (2024) · Alexandria Engineering Journal · doi:10.1016/j.aej.2024.10.039 · Cited by 7
Caponio, C., Stano, P., Carli, R., Olivieri, I., Ragone, D., Sorniotti, A., Gruber, P., & Montanaro, U. (2024) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2024.3513701 · Cited by 7
“Such degradation in experimental closed-loop following performance can be attributed to mismatches between the simulation environment used for the training phase and the experimental one, highlighting a drawback of DRL known as the simulation-to-reality gap [19].”
Integrating Reinforcement Learning and Virtual Fixtures for Safer Automatic Robotic Surgery
Fan, K., & Chen, Z. (2025) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2025.3559826 · Cited by 5
“Another challenge is transferring the policy to real robots [6].”
Isaac Sim-to-Real: Reinforcement Learning based Locomotion for Quadrupeds
Dowdy, J., & Vaz, J. C. (2025) · IEEE International Conference on Automation Science and Engineering (CASE) · doi:10.1109/case58245.2025.11163761 · Cited by 9
“However, despite encouraging simulation results, the well-known sim-to-real gap presents significant challenges when implementing these policies on physical robots [12].”
Route Optimization for UGVs: A Systematic Analysis of Applications, Algorithms and Challenges
Ponce, D. F. Y., Montalvo, W., Guamán-Gavilanes, X. A., & Echeverría-Cadena, M. D. (2025) · Applied Sciences · doi:10.3390/app15126477 · Cited by 5
“This collaborative approach is vital for transforming promising lab-scale proofs-of-concept into robust and fieldable solutions, thereby bridging the gap between theoretical developments and practical deployment [63].”
Mobile Robot Control and Autonomy Through Collaborative Twin
Tahir, N., & Parasuraman, R. N. (2023) · IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops) · doi:10.1109/percomworkshops56833.2023.10150325 · Cited by 8
“This paradigm also involves robot control in a physically secure and technologically adaptable space while bridging the mismatch between simulated and real environment, thus helping analyze the Simulation to Reality Gap [8].”
BackgroundSafe and Robust Simulation-to-Reality Transfer for Mobile Robots
Valdivia, J. P., Hata, A., & Terra, A. (2024) · IEEE International Conference on Emerging Technologies and Factory Automation (ETFA) · doi:10.1109/etfa61755.2024.10711026 · Cited by 5
“In transfer learning (TL) and more specifically, fine-tuning, a model f ( θ ) (where θ represents its parameters) that was trained to accomplish a task T A is adapted to perform another related task T B [11].”
BackgroundMethodOtto—Design and Control of an 8-DoF SEA-Driven Quadrupedal Robot
Scaldaferri, A., Tolomei, S., Iotti, F., Gambino, P., Pierallini, M., Angelini, F., & Garabini, M. (2025) · IEEE Open Journal of the Industrial Electronics Society · doi:10.1109/ojies.2025.3567112 · Cited by 5
“We employ different techniques to close the Sim2Real gap [59], [60], [61], [62] and successfully deploy the trained policy on the real robot.”
Wen, Y., Wan, Z., Zhou, M. Q., Hou, S., Cao, Z., Le, C., Chen, J., Tian, Z., Zhang, W., & Wang, Y.-X. (2023) · CAAI Artificial Intelligence Research · doi:10.26599/air.2023.9150026 · Cited by 16
A framework for process anomaly detection in 3D concrete printing
Kamhawi, A., Lin, Y., Watson, C., Barton, K. L., & Meibodi, M. A. (2025) · Additive manufacturing · doi:10.1016/j.addma.2025.105056 · Cited by 6
Zhou, Z., Lu, Y., Tortós-Vinocour, P. E., Qin, R., Kokubu, S., Matsunaga, F., Xie, Q., & Yu, W. (2024) · Frontiers in Bioengineering and Biotechnology · doi:10.3389/fbioe.2024.1334643 · Cited by 4
“To bridge the gap between simulated and real-world environments with imbalanced data, researchers have proposed some methods, which are mainly categorized into domain randomization and domain adaptation (Salvato et al., 2021).”
HiER: Highlight Experience Replay for Boosting Off-Policy Reinforcement Learning Agents
Horváth, D., Martín, J. B., Erdős, G., Istenes, Z., & Moutarde, F. (2024) · IEEE Access · doi:10.1109/access.2024.3427012 · Cited by 5
“While transferring knowledge [1], [2] between domains [3], [4], [5], [6], robotic systems [7], or tasks [8] is fundamental, it is essential to create and apply universal methods such as reinforcement-learning-based algorithms (RL) [9], [10], [11] which are inspired by the profoundly universal…”
BackgroundPadalkar, A., Quere, G., Raffin, A., Silvério, J., & Stulp, F. (2024) · Autonomous Robots · doi:10.1007/s10514-024-10164-6 · Cited by 12
“Simulation and sim-to-real methods resolve these issues, but come with their own set of challenges when simulating in-contact tasks (Salvato et al., 2021).”
BackgroundReinforcement learning and the Metaverse: a symbiotic collaboration
Elsokkary, N., Khan, W., Shurrab, M., Mizouni, R., Singh, S., Bentahar, J., Mourad, A., & Otrok, H. (2025) · Artificial Intelligence Review · doi:10.1007/s10462-025-11433-1 · Cited by 8
Fu, Y., Türkcan, M. K., Ghasemi, M., Mo, Z., Zang, C., Adhikari, A., Kostić, Z., Zussman, G., & Di, X. S. (2026) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2026.3656165 · Cited by 24
“Although these gaps penetrate through each stage along the DT pipeline, from computer vision, tracking, to prediction, and intervention, sim2real transfer is more studied in object detection [217] and policy learning [218], [219], [220], [221].”
Hathaway, J., Stolkin, R., & Rastegarpanah, A. (2024) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros58592.2024.10802660 · Cited by 4
“Previous approaches to sim to real adaptation of can be broadly categorised into domain adaptive and transfer learning approaches [7].”
Sun, L., & Yoshiyasu, Y. (2025) · Image and Vision Computing · doi:10.1016/j.imavis.2025.105522 · Cited by 8
Fundamentals of Evolutionary Machine Learning
Banzhaf, W., & Machado, P. (2023) · Genetic and evolutionary computation · doi:10.1007/978-981-99-3814-8_1 · Cited by 7
Coupled Requirements-Driven Testing of CPS: From Simulation to Reality
Agrawal, A., Zech, P., & Vierhauser, M. (2024) · IEEE International Requirements Engineering Conference (RE) · doi:10.1109/re59067.2024.00040 · Cited by 4
Mortensen, A. B., Pedersen, E., Benedicto, L. V., Burg, L., Madsen, M. R., & Bøgh, S. (2024) · International Conference on Space Robotics (iSpaRo) · doi:10.1109/isparo60631.2024.10687660 · Cited by 3
“Salvato et al.[16] introduce adversarial RL and transfer learning as alternative strategies.”
BackgroundModel-free robust reinforcement learning via Polynomial Chaos
Liu, J., Wu, F., & Zhang, X. (2024) · Knowledge-Based Systems · doi:10.1016/j.knosys.2024.112783 · Cited by 4
Scenario-Based Field Testing of Drone Missions
Vierhauser, M., Meixner, K., & Biffl, S. (2024) · Euromicro Conference on Software Engineering and Advanced Applications (SEAA) · doi:10.1109/seaa64295.2024.00012 · Cited by 7
“This simulator-to-reality gap has been recognized as a critical issue in robotics research [5], describing the mismatch of simulated and real physical-law performances caused by the inaccurate representation of the real environment in simulation [6].”
Impact of Morphology Variations on Evolved Neural Controllers for Modular Robots
Medvet, E., & Rusin, F. (2023) · Communications in computer and information science · doi:10.1007/978-3-031-31183-3_22 · Cited by 3
Self-citationLearning to Navigate Through Reinforcement Across the Sim2Real GapPreprint
Azzam, R., Chehadeh, M. S., Hay, O. A., Boiko, I. M., & Zweiri, Y. (2022) · doi:10.36227/techrxiv.20138960.v1 · Cited by 3
Hydraulic-Supports Alignment by TD3 with Segmented Experience Pool
Yang, Y., Dai, Y., Wang, T., & Qian, W. (2025) · Neural Processing Letters · doi:10.1007/s11063-025-11744-y · Cited by 2
Learning Speed-Adaptive Walking Agent Using Imitation Learning with Physics-Informed Simulation
Chiu, Y.-H., Lee, U. H., Song, C.-S., Hu, M., & Kang, I. (2025) · International Conference on Rehabilitation Robotics · doi:10.1109/icorr66766.2025.11063216 · Cited by 8
“However, a significant challenge remains in ensuring that these digital twins accurately represent the real human body and its interactions with the environment, which is commonly referred as a sim-to-real gap [14], [15].”
Multi-Robot Collaboration Through Reinforcement Learning and Abstract Simulation
Labiosa, A., & Hanna, J. P. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11127323 · Cited by 3
“Therefore, many works adopt the sim2real paradigm – training policies in simulation and transferring the fixed policies to physical robots [34], [35].”
Guiding Reinforcement Learning with Shared Control Templates
Padalkar, A., Quere, G., Steinmetz, F., Raffin, A., Nieuwenhuisen, M., Silvério, J., & Stulp, F. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10161058 · Cited by 4
“In summary, RL-SCT leverages pre-specified task knowledge to improve sample efficiency and simplify reward function design, facilitating direct learning on real robots and hence alleviating the need for simulation, a well-known problem in RL [20], [21].”
MethodDrone-racing champions outpaced by AI
de Croon, G. (2023) · Nature · doi:10.1038/d41586-023-02506-8 · Cited by 3
Intervention-Assisted Online Deep Reinforcement Learning for Stochastic Queuing Network Optimization
Wigmore, J., Shrader, B., & Modiano, E. (2024) · ACM International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing (MobiHoc) · doi:10.1145/3641512.3686383 · Cited by 5
“This is often referred to as the sim-to-real gap [13, 21].”
Achieving the Safety and Security of the End-to-End AV Pipeline
Curran, N. T., Cho, M., Feng, R., Liu, L., Tang, B., MohajerAnsari, P., Domeke, A., Pesé, M. D., & Shin, K. G. (2023) · doi:10.1145/3689936.3694694 · Cited by 3
Cross-Platform Learning-Based Fault Tolerant Surfacing Controller for Underwater Robots
Hamamatsu, Y., Remmas, W., Rebane, J., Kruusmaa, M., & Ristolainen, A. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11127355 · Cited by 3
“Furthermore, to illustrate the practical applicability, we conducted Sim2Real transfer experiments with U-CAT in a swimming pool, successfully deploying the controller trained entirely in simulation in real-world scenarios [17], [18], [19], [20].”
An environment emulator for training a neural network model to solve the “Following the leader” task
Selivanov, A., Борисович, Р. Р., Gryaznov, A. V., Shein, V. A., & Sboev, A. G. (2022) · Procedia Computer Science · doi:10.1016/j.procs.2022.11.058 · Cited by 2
Nutzung von KI-Methoden für die Kupplungsentwicklung in automobilen Antriebssträngen
Altenburg, S., Schuchter, F., Bause, K., & Albers, A. (2023) · Forschung im Ingenieurwesen · doi:10.1007/s10010-023-00665-8 · Cited by 2
“Daher werden DRL-Trainings meistens mit einer simulativen Umgebung durchgeführt und gute Agenten auf das zu regelnde System übertragen [29].”
Background
Metadata
- Authors
- Salvato, E., Fenu, G., Medvet, E., & Pellegrino, F. A.
- Affiliations
- University of Trieste (IT)
- Venue
- IEEE Access
- Published
- 2021
- Volume
- 9
- Pages
- 153171-153187
- ISSN
- 2169-3536 (ISSN-L)
- OpenAlex
- W3212580602
- Cited by
- 200
- References
- 173
- Citations by year
- 2026: 45 · 2025: 51 · 2024: 53 · 2023: 39 · 2022: 12
- FWCI
- 9.15
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- IEEE Access (publishedVersion, cc-by)
- ArTS Archivio della ricerca di Trieste (University of Trieste https://www.units.it/) (submittedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Italian Ministry for Research in the framework of the 2017 Program for Research Projects of National Interest (2017YKXYXJ)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Physical Sciences › Engineering › Control and Systems Engineering › Robot Manipulation and Learning
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Keywords
- sim-to-real transfer, reality gap, robot control, reinforcement learning, robot simulator, robot dynamics modeling
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
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.