{"title":"非线性串联弹性作动器控制的模型反演方法","authors":"Christopher Jarrett, A. McDaid","doi":"10.1109/ICORR.2019.8779470","DOIUrl":null,"url":null,"abstract":"This paper presents a model inversion procedure for a viscoelastic compliant element contained within a rotary series elastic actuator (SEA). Model inversion plays an important role in enabling accurate model-based control of physical human-robot interaction (HRI) with SEAs. If the compliant element of the SEA is elastomeric, analytically inverting its model is non-trivial due to the presence of complex non-linear terms. This paper applies an alternative inversion procedure, by coupling a partially analytical inverse model with a disturbance observer (DOB). Results of inverting without a DOB are given as a baseline and compared to the analytical inversion + DOB with two different types of filter. To quantify the accuracy of the inversion, the output of the inversion procedure is passed back through the forward model to identify the ‘actual setpoint’, the signal that would be generated if the output of the inversion procedure was tracked accurately. The inversions for five desired torque signals are presented; in all cases, the root-mean square (RMS) error between the desired setpoint and actual setpoint is lower when the DOB inversion procedure is used, compared to the RMS error incurred when the DOB is omitted. The results suggest that the proposed inversion procedure provides an accurate and mathematically tractable inverse of the complex viscoelastic elastomer model.","PeriodicalId":130415,"journal":{"name":"2019 IEEE 16th International Conference on Rehabilitation Robotics (ICORR)","volume":"84 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2019-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":"{\"title\":\"A Model Inversion Procedure for Control of Nonlinear Series Elastic Actuators\",\"authors\":\"Christopher Jarrett, A. McDaid\",\"doi\":\"10.1109/ICORR.2019.8779470\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"This paper presents a model inversion procedure for a viscoelastic compliant element contained within a rotary series elastic actuator (SEA). Model inversion plays an important role in enabling accurate model-based control of physical human-robot interaction (HRI) with SEAs. If the compliant element of the SEA is elastomeric, analytically inverting its model is non-trivial due to the presence of complex non-linear terms. This paper applies an alternative inversion procedure, by coupling a partially analytical inverse model with a disturbance observer (DOB). Results of inverting without a DOB are given as a baseline and compared to the analytical inversion + DOB with two different types of filter. To quantify the accuracy of the inversion, the output of the inversion procedure is passed back through the forward model to identify the ‘actual setpoint’, the signal that would be generated if the output of the inversion procedure was tracked accurately. The inversions for five desired torque signals are presented; in all cases, the root-mean square (RMS) error between the desired setpoint and actual setpoint is lower when the DOB inversion procedure is used, compared to the RMS error incurred when the DOB is omitted. The results suggest that the proposed inversion procedure provides an accurate and mathematically tractable inverse of the complex viscoelastic elastomer model.\",\"PeriodicalId\":130415,\"journal\":{\"name\":\"2019 IEEE 16th International Conference on Rehabilitation Robotics (ICORR)\",\"volume\":\"84 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2019-06-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"2\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2019 IEEE 16th International Conference on Rehabilitation Robotics (ICORR)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ICORR.2019.8779470\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2019 IEEE 16th International Conference on Rehabilitation Robotics (ICORR)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICORR.2019.8779470","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
A Model Inversion Procedure for Control of Nonlinear Series Elastic Actuators
This paper presents a model inversion procedure for a viscoelastic compliant element contained within a rotary series elastic actuator (SEA). Model inversion plays an important role in enabling accurate model-based control of physical human-robot interaction (HRI) with SEAs. If the compliant element of the SEA is elastomeric, analytically inverting its model is non-trivial due to the presence of complex non-linear terms. This paper applies an alternative inversion procedure, by coupling a partially analytical inverse model with a disturbance observer (DOB). Results of inverting without a DOB are given as a baseline and compared to the analytical inversion + DOB with two different types of filter. To quantify the accuracy of the inversion, the output of the inversion procedure is passed back through the forward model to identify the ‘actual setpoint’, the signal that would be generated if the output of the inversion procedure was tracked accurately. The inversions for five desired torque signals are presented; in all cases, the root-mean square (RMS) error between the desired setpoint and actual setpoint is lower when the DOB inversion procedure is used, compared to the RMS error incurred when the DOB is omitted. The results suggest that the proposed inversion procedure provides an accurate and mathematically tractable inverse of the complex viscoelastic elastomer model.