File Download
Links for fulltext
(May Require Subscription)
- Publisher Website: 10.1109/ACCESS.2021.3106327
- Scopus: eid_2-s2.0-85113314406
- WOS: WOS:000690437300001
Supplementary
- Citations:
- Appears in Collections:
Article: Tactile Servoing Based Pressure Distribution Control of a Manipulator Using a Convolutional Neural Network
Title | Tactile Servoing Based Pressure Distribution Control of a Manipulator Using a Convolutional Neural Network |
---|---|
Authors | |
Keywords | Manipulator control Pressure distribution control Tactile servoing Convolutional neural network |
Issue Date | 2021 |
Citation | IEEE Access, 2021, v. 9, p. 117132-117139 How to Cite? |
Abstract | In this paper, we propose a novel tactile servoing based pressure distribution control scheme of a manipulator using a convolutional neural network (CNN). The CNN significantly improves the performance of the tactile servoing scheme compared to the one based on the tactile Jacobian. LeNet-5, originally proposed for image classification problems, is applied to represent a nonlinear relationship between current and desired pressure distributions and the robot velocity command by using mean squared error as the loss function. In the proposed control scheme, the trained CNN directly generates the velocity command of the manipulator so that the pressure distribution converges to a given desired pressure distribution. Validation experiments are carried out to evaluate the performance of the proposed control scheme. Experimental results show that the proposed tactile servoing control scheme has better performance than the Jacobian-based tactile servoing control scheme. |
Persistent Identifier | http://hdl.handle.net/10722/303043 |
ISI Accession Number ID |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Wen, Chen-Ting | - |
dc.contributor.author | Arai, Shogo | - |
dc.contributor.author | Kinugawa, Jun | - |
dc.contributor.author | Kosuge, Kazuhiro | - |
dc.date.accessioned | 2021-09-07T08:43:05Z | - |
dc.date.available | 2021-09-07T08:43:05Z | - |
dc.date.issued | 2021 | - |
dc.identifier.citation | IEEE Access, 2021, v. 9, p. 117132-117139 | - |
dc.identifier.uri | http://hdl.handle.net/10722/303043 | - |
dc.description.abstract | In this paper, we propose a novel tactile servoing based pressure distribution control scheme of a manipulator using a convolutional neural network (CNN). The CNN significantly improves the performance of the tactile servoing scheme compared to the one based on the tactile Jacobian. LeNet-5, originally proposed for image classification problems, is applied to represent a nonlinear relationship between current and desired pressure distributions and the robot velocity command by using mean squared error as the loss function. In the proposed control scheme, the trained CNN directly generates the velocity command of the manipulator so that the pressure distribution converges to a given desired pressure distribution. Validation experiments are carried out to evaluate the performance of the proposed control scheme. Experimental results show that the proposed tactile servoing control scheme has better performance than the Jacobian-based tactile servoing control scheme. | - |
dc.language | eng | - |
dc.relation.ispartof | IEEE Access | - |
dc.rights | This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. | - |
dc.subject | Manipulator control | - |
dc.subject | Pressure distribution control | - |
dc.subject | Tactile servoing | - |
dc.subject | Convolutional neural network | - |
dc.title | Tactile Servoing Based Pressure Distribution Control of a Manipulator Using a Convolutional Neural Network | - |
dc.type | Article | - |
dc.description.nature | published_or_final_version | - |
dc.identifier.doi | 10.1109/ACCESS.2021.3106327 | - |
dc.identifier.scopus | eid_2-s2.0-85113314406 | - |
dc.identifier.hkuros | 328271 | - |
dc.identifier.volume | 9 | - |
dc.identifier.spage | 117132 | - |
dc.identifier.epage | 117139 | - |
dc.identifier.eissn | 2169-3536 | - |
dc.identifier.isi | WOS:000690437300001 | - |