A Novel Adaptive Controller for Robot Manipulators Based on Active Inference
Open accessThe DOI matches its Crossref record; 27 of the 28 references it cites were found in a registry or an index.
Abstract
More adaptive controllers for robot manipulators are needed, which can deal with large model uncertainties. This letter presents a novel active inference controller (AIC) as an adaptive control scheme for industrial robots. This scheme is easily scalable to high degrees-of-freedom, and it maintains high performance even in the presence of large unmodeled dynamics. The proposed method is based on active inference, a promising neuroscientific theory of the brain, which describes a biologically plausible algorithm for perception and action. In this work, we formulate active inference from a control perspective, deriving a model-free control law which is less sensitive to unmodeled dynamics. The performance and the adaptive properties of the algorithm are compared to a state-of-the-art model reference adaptive controller (MRAC) in an experimental setup with a real7-DOF robot arm. The results showed that the AIC outperformed the MRAC in terms of adaptability, providing a more general control law. This confirmed the relevance of active inference for robot control.
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 says what the claim attributes to it.
In the citing paper
“Active inference can unify 1) abstract decision-making with guarantees, as in this article, 2) adaptive [12] and fault tolerant [10], [11], [46] torque control, as well as 3) state estimation and learning.”
From Active Inference and Behavior Trees for Reactive Action Planning and Execution in Robotics (2023) · cites this paper as [12]
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
The free-energy principle: a unified brain theory?
Friston, K. (2010) · Nature Reviews Neuroscience · doi:10.1038/nrn2787 · Cited by 8,120 · cited 14×
“The proposed control scheme is based on the general freeenergy principle proposed by Karl Friston [7], and redefined in engineering terms [8], [9].”
BackgroundInfluentialHierarchical Models in the Brain
Friston, K. (2008) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1000211 · Cited by 952 · cited 1×
“1Generalised motions can extend up to infinite order but the noise at high orders is predominant, thus we can limit the chosen order to nd [26].”
BackgroundVariational free energy and the Laplace approximation
Friston, K., Mattout, J., Trujillo-Barreto, N., Ashburner, J., & Penny, W. (2007) · NeuroImage · doi:10.1016/j.neuroimage.2006.08.035 · Cited by 880 · cited 3×
“More precisely, the generalised motions μ̃ of the belief under local linearity assumptions [24] are, up to the second order:”
BackgroundAction and behavior: a free-energy formulation
Friston, K. J., Daunizeau, J., Kilner, J., & Kiebel, S. J. (2010) · Biological Cybernetics · doi:10.1007/s00422-010-0364-z · Cited by 866 · cited 1×
“In this view, the motor control of human body can be considered as the fulfillment of a prior expectation about proprioceptive sensations [11].”
BackgroundTheory and applications of adaptive control—A survey
Åström, K. J. (1983) · Automatica · doi:10.1016/0005-1098(83)90002-x · Cited by 748 · cited 1×
“On the other hand, the adaptive control branch of control theory [17], offers solutions to deal with manipulators subject to parameters variation and abrupt changes in the dynamics.”
BackgroundAction understanding and active inference
Friston, K., Mattout, J., & Kilner, J. (2011) · Biological Cybernetics · doi:10.1007/s00422-011-0424-z · Cited by 665 · cited 2×
“The proposed control scheme is based on the general freeenergy principle proposed by Karl Friston [7], and redefined in engineering terms [8], [9].”
Method1. Bayesian Statistics a Review
Lindley, D. V. (1972) · Bayesian Statistics · doi:10.1137/1.9781611970654.ch1 · Cited by 521
Reinforcement learning or active inference?
Friston, K. J., Daunizeau, J., & Kiebel, S. J. (2009) · PLoS ONE · doi:10.1371/journal.pone.0006421 · Cited by 443 · cited 1×
“The next two subsections describe the approach proposed by Friston [10], [27] to minimise F , using gradient descent.”
The free energy principle for action and perception: A mathematical review
Buckley, C. L., Kim, C. S., McGregor, S., & Seth, A. K. (2017) · Journal of Mathematical Psychology · doi:10.1016/j.jmp.2017.09.004 · Cited by 363 · cited 10×
“Assumption 3: The Gaussian noise affecting the different sensory channels is supposed uncorrelated [8], [24].”
BackgroundMethodInfluentialAdaptive linear controller for robotic manipulators
Koivo, A., & Guo, T.-H. (1983) · IEEE Transactions on Automatic Control · doi:10.1109/tac.1983.1103211 · Cited by 331 · cited 1×
“The other most common approach for robot control is the self-tuning adaptive control [21], [22].”
BackgroundA tutorial on the free-energy framework for modelling perception and learning
Bogacz, R. (2017) · Journal of Mathematical Psychology · doi:10.1016/j.jmp.2015.11.003 · Cited by 283 · cited 1×
“The proposed control scheme is based on the general freeenergy principle proposed by Karl Friston [7], and redefined in engineering terms [8], [9].”
MethodA review on model reference adaptive control of robotic manipulators
Zhang, D., & Wei, B. (2017) · Annual Reviews in Control · doi:10.1016/j.arcontrol.2017.02.002 · Cited by 208 · cited 1×
“The first technique being studied for robot manipulators was the model reference adaptive control (MRAC) [19].”
MethodAdaptive control of robot manipulators - A review
Hsia, T. (1986) · Proceedings. 1986 IEEE International Conference on Robotics and Automation · doi:10.1109/robot.1986.1087696 · Cited by 176 · cited 1×
“Within adaptive controllers, two main categories can be identified: the model reference adaptive systems, and the self-tuning regulators [18].”
BackgroundFriston, K., Stephan, K., Li, B., & Daunizeau, J. (2010) · Mathematical Problems in Engineering · doi:10.1155/2010/621670 · Cited by 171 · cited 2×
“The belief update law for state estimation is determined from the gradient of the free-energy, with respect to each generalised motion [8], [25]:”
BackgroundLocally Weighted Projection Regression: Incremental Real Time Learning in High Dimensional Space
Vijayakumar, S., & Schaal, S. (2000) · International Conference on Machine Learning · Cited by 95
Active inference and robot control: a case study
Pio-Lopez, L., Nizard, A., Friston, K., & Pezzulo, G. (2016) · Journal of The Royal Society Interface · doi:10.1098/rsif.2016.0616 · Cited by 84 · cited 4×
“Even if the neuronal simulations are a strong proof of concept for the neuroscientific theory, in the present form their extension to realistic robotic scenarios [12], [13] does not provide advantages over other classical controllers.”
BackgroundMethodInfluentialAdaptive Robot Body Learning and Estimation Through Predictive Coding
Lanillos, P., & Cheng, G. (2018) · 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8593684 · Cited by 71 · cited 1×
“Another recent work [14] formalised the use of the free-energy for static state estimation, using a real UR5 robot arm equipped with proprioceptive and visual sensors.”
MethodLearning Robot Dynamics for Computed Torque Control Using Local Gaussian Processes Regression
Nguyen-Tuong, D., & Peters, J. (2008) · ECSIS Symposium on Learning and Adaptive Behaviors for Robotic Systems (LAB-RS) · doi:10.1109/lab-rs.2008.16 · Cited by 53 · cited 2×
“Another optimised computed torque algorithm such as LGP [28], which relies on learning dynamical models, has a cost ofO(N2) for online learning, where N is the number of data points (i.”
BackgroundLedezma, F. D., & Haddadin, S. (2017) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2017.8246910 · Cited by 32 · cited 1×
“Even though it is possible to exploit the physical knowledge of the system to simplify and improve the learning performance [4], the need of large amount”
BackgroundIncremental learning of context-dependent dynamic internal models for robot control
Jamone, L., Damas, B., & Santos-Victor, J. (2014) · The proceedings of the IEEE Conference on Control Applications/The proceedings of the ... IEEE Conference on Control Applications · doi:10.1109/isic.2014.6967617 · Cited by 30 · cited 1×
“2974451 of training data and several iterations for learning, still remains a problem and hard to generalise [5], [6].”
BackgroundActive inference body perception and action for humanoid robots
Oliver, G., Lanillos, P., & Cheng, G. (2019) · arXiv · doi:10.48550/arxiv.1906.03022 · Cited by 30 · cited 1×
“In a parallel, related work on active inference [16], the authors successfully controlled a real 3-DOF robot arm using velocity commands.”
BackgroundA new data source for inverse dynamics learning
Kappler, D., Meier, F., Ratliff, N., & Schaal, S. (2017) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2017.8206345 · Cited by 25 · cited 1×
“2974451 of training data and several iterations for learning, still remains a problem and hard to generalise [5], [6].”
BackgroundApplication of a self-tuning pole-placement regulator to an industrial manipulator
Walters, R. C., & Bayoumi, M. M. (1982) · IEEE Conference on Decision and Control (CDC) · doi:10.1109/cdc.1982.268454 · Cited by 22 · cited 1×
“The other most common approach for robot control is the self-tuning adaptive control [21], [22].”
BackgroundActive inference with function learning for robot body perception
Lanillos, P., & Cheng, G. (2018) · mediaTUM (Technical University of Munich) · Cited by 13
ELeaRNT: Evolutionary Learning of Rich Neural Network Topologies
Matteucci, M. (2006) · Defense Technical Information Center (DTIC) · doi:10.21236/ada456062 · Cited by 8 · cited 1×
“chine learning methods to obtain accurate inverse dynamic models [1,2]. In general, learning models using Neural Networks (NN) requires experts for defining the best topology for a particular problem [3]. Even though it is possible to exploit the physical knowledge of the system to simplify and improve the learning pe…”
MethodHyperstability approach to the synthesis of adaptive controllers for robot manipulators
Tarokh, M. (2002) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/robot.1991.131947 · Cited by 6 · cited 5×
“The control architecture is taken from [20], where the control is specified in terms of feedforward and feedback adaptive gain matrices.”
BackgroundMethodInfluentialRobot manipulator control under the Active Inference framework
Mercadé, A. C. (2018) · Cited by 2
Research works printed but not linked: 1
- [2] D. Nguyen-Tuong, J. Peters, and M. Seeger, 'Local gaussian process regression for real time online model learning,' in Proc of Neural Information Processing Systems (NIPS2008) , 2008, pp. 1193-1200. -
Cited by
Predictive Coding: Towards a Future of Deep Learning beyond Backpropagation?
Millidge, B., Salvatori, T., Song, Y., Bogacz, R., & Lukasiewicz, T. (2022) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2022/774 · Cited by 91
An Empirical Study of Active Inference on a Humanoid Robot
Oliver, G., Lanillos, P., & Cheng, G. (2022) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2021.3049907 · Cited by 63
“Recently, an active inference controller, working in the joint space for an industrial 7-DOF robot arm, was tested and compared to an adaptive controller, obtaining compelling results [20].”
BackgroundMethodResultBayesian mechanics for stationary processes
Da Costa, L., Friston, K., Heins, C., & Pavliotis, G. A. (2021) · Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences · doi:10.1098/rspa.2021.0518 · Cited by 63
“Active inference is used in various domains to simulate control [65,69,71,72,74–77], thus, it is natural that we can relate the dynamics of active states to well-known forms of stochastic control.”
BackgroundFixed-Time Fuzzy Control of Uncertain Robots With Guaranteed Transient Performance
Zhu, C., Yang, C., Jiang, Y., & Zhang, H. (2022) · IEEE Transactions on Fuzzy Systems · doi:10.1109/tfuzz.2022.3194373 · Cited by 61
“In [10], a novel adaptive controller based on active inference has been proposed for the robot to handle large model uncertainties and a large number DOFs, which enable to deal with the robot uncertainty, besides the NN and the FLS.”
BackgroundPredictive processing in cognitive robotics: A review
Ciria, A., Schillaci, G., Pezzulo, G., Hafner, V. V., & Lara, B. (2021) · Neural Computation · doi:10.1162/neco_a_01383 · Cited by 51
“Similarly, Pezzato et al. (2020) present an active inference framework using a pre-coded controller and a generative function.”
Method
Metadata
- Authors
- Pezzato, C., Ferrari, R., & Corbato, C. H.
- Affiliations
- Delft University of Technology (NL)
- Venue
- IEEE Robotics and Automation Letters
- Published
- 2020-04
- Volume
- 5
- Issue
- 2
- Pages
- 2973-2980
- ISSN
- 2377-3766 (ISSN-L)
- 2377-3774
- OpenAlex
- W2976738340
- Cited by
- 104
- References
- 28
- Citations by year
- 2026: 19 · 2025: 14 · 2024: 17 · 2023: 15 · 2022: 15 · 2021: 20 · 2020: 4
- FWCI
- 6.68
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Robotics and Automation Letters (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Research Repository (Delft University of Technology) (submittedVersion)
- Data Archiving and Networked Services (DANS) (submittedVersion, other-oa)
- arXiv (Cornell University)
- Funders
- Ahold Delhaize
- Topics
- Physical Sciences › Engineering › Control and Systems Engineering › Adaptive Control of Nonlinear Systems
- Life Sciences › Neuroscience › Cognitive Neuroscience › Embodied and Extended Cognition
- Life Sciences › Neuroscience › Cognitive Neuroscience › Motor Control and Adaptation
- Keywords
- active inference, adaptive control, robot manipulators, model-free control, unmodeled dynamics, model uncertainty, model reference adaptive control, 7-DOF robot arm, industrial robots, high degrees of freedom
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.