{"title":"Data-driven inverse dynamics modeling using neural-networks and regression-based techniques","authors":"Maciej Pikuliński, Paweł Malczyk, Ronald Aarts","doi":"10.1007/s11044-024-10024-2","DOIUrl":null,"url":null,"abstract":"<p>This research proposes a novel approach for the residual modeling of inverse dynamics employed to control a real robotic device. Specifically, we use techniques based on linear regression for residual modeling while a nominal model is discovered by physics-informed neural networks such as the Lagrangian Neural Network and the Feedforward Neural Network. We introduce an efficient online learning mechanism for the residual models that utilizes rank-one updates based on the Sherman–Morrison formula. This enables faster adaptation and updates to effects not captured by the neural networks. While the time complexity of updating the model is comparable to other successful learning methods, the method excels in prediction complexity, which depends solely on the model dimension. We propose two online learning strategies: a weighted approach that gradually diminishes the influence of past measurements on the model, and a windowed approach that sharply excludes the oldest data from impacting the model. We explore the relationship between these strategies, offering recommendations for parameter selection and practical application. Special attention is given to optimizing the computation time of the weighted approach when recomputation techniques are implemented, which results in comparable or even lower execution times of the weighted controller than the windowed one. Additionally, we assess other methods, such as the Woodbury identity, QR decomposition, and Cholesky decomposition, which can be implicitly used to update the model. We empirically validate our approach using real data from a 2-degrees-of-freedom flexible manipulator, demonstrating consistent improvements in feedforward controller performance.</p>","PeriodicalId":49792,"journal":{"name":"Multibody System Dynamics","volume":null,"pages":null},"PeriodicalIF":2.6000,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Multibody System Dynamics","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.1007/s11044-024-10024-2","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"MECHANICS","Score":null,"Total":0}
引用次数: 0
Abstract
This research proposes a novel approach for the residual modeling of inverse dynamics employed to control a real robotic device. Specifically, we use techniques based on linear regression for residual modeling while a nominal model is discovered by physics-informed neural networks such as the Lagrangian Neural Network and the Feedforward Neural Network. We introduce an efficient online learning mechanism for the residual models that utilizes rank-one updates based on the Sherman–Morrison formula. This enables faster adaptation and updates to effects not captured by the neural networks. While the time complexity of updating the model is comparable to other successful learning methods, the method excels in prediction complexity, which depends solely on the model dimension. We propose two online learning strategies: a weighted approach that gradually diminishes the influence of past measurements on the model, and a windowed approach that sharply excludes the oldest data from impacting the model. We explore the relationship between these strategies, offering recommendations for parameter selection and practical application. Special attention is given to optimizing the computation time of the weighted approach when recomputation techniques are implemented, which results in comparable or even lower execution times of the weighted controller than the windowed one. Additionally, we assess other methods, such as the Woodbury identity, QR decomposition, and Cholesky decomposition, which can be implicitly used to update the model. We empirically validate our approach using real data from a 2-degrees-of-freedom flexible manipulator, demonstrating consistent improvements in feedforward controller performance.
期刊介绍:
The journal Multibody System Dynamics treats theoretical and computational methods in rigid and flexible multibody systems, their application, and the experimental procedures used to validate the theoretical foundations.
The research reported addresses computational and experimental aspects and their application to classical and emerging fields in science and technology. Both development and application aspects of multibody dynamics are relevant, in particular in the fields of control, optimization, real-time simulation, parallel computation, workspace and path planning, reliability, and durability. The journal also publishes articles covering application fields such as vehicle dynamics, aerospace technology, robotics and mechatronics, machine dynamics, crashworthiness, biomechanics, artificial intelligence, and system identification if they involve or contribute to the field of Multibody System Dynamics.