Abstract
We propose a practical state-of-the-art method to develop a machine-learning-based humanoid robot that can work as a production line worker. The proposed approach provides an intuitive way to collect data and exhibits the following characteristics: task performing capability, task reiteration ability, generalizability, and easy applicability. The proposed approach utilizes a real-time user interface with a monitor and provides a first-person perspective using a head-mounted display. Through this interface, teleoperation is used for collecting task operating data, especially for tasks that are difficult to be applied with a conventional method. A two-phase deep learning model is also utilized in the proposed approach. A deep convolutional autoencoder extracts images features and reconstructs images, and a fully connected deep time delay neural network learns the dynamics of a robot task process from the extracted image features and motion angle signals. The 'Nextage Open' humanoid robot is used as an experimental platform to evaluate the proposed model. The object folding task utilizing with 35 trained and 5 untrained sensory motor sequences for test. Testing the trained model with online generation demonstrates a 77.8% success rate for the object folding task.
| Original language | English |
|---|---|
| Article number | 7762066 |
| Pages (from-to) | 397-403 |
| Number of pages | 7 |
| Journal | IEEE Robotics and Automation Letters |
| Volume | 2 |
| Issue number | 2 |
| DOIs | |
| Publication status | Published - 2017 Apr 1 |
Fingerprint
Keywords
- Humanoid robots
- learning and adaptive systems
- motion control of manipulators
- neurorobotics
ASJC Scopus subject areas
- Control and Systems Engineering
- Human-Computer Interaction
- Biomedical Engineering
- Mechanical Engineering
- Control and Optimization
- Artificial Intelligence
- Computer Science Applications
- Computer Vision and Pattern Recognition
Cite this
Repeatable Folding Task by Humanoid Robot Worker Using Deep Learning. / Yang, Pin Chu; Sasaki, Kazuma; Suzuki, Kanata; Kase, Kei; Sugano, Shigeki; Ogata, Tetsuya.
In: IEEE Robotics and Automation Letters, Vol. 2, No. 2, 7762066, 01.04.2017, p. 397-403.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - Repeatable Folding Task by Humanoid Robot Worker Using Deep Learning
AU - Yang, Pin Chu
AU - Sasaki, Kazuma
AU - Suzuki, Kanata
AU - Kase, Kei
AU - Sugano, Shigeki
AU - Ogata, Tetsuya
PY - 2017/4/1
Y1 - 2017/4/1
N2 - We propose a practical state-of-the-art method to develop a machine-learning-based humanoid robot that can work as a production line worker. The proposed approach provides an intuitive way to collect data and exhibits the following characteristics: task performing capability, task reiteration ability, generalizability, and easy applicability. The proposed approach utilizes a real-time user interface with a monitor and provides a first-person perspective using a head-mounted display. Through this interface, teleoperation is used for collecting task operating data, especially for tasks that are difficult to be applied with a conventional method. A two-phase deep learning model is also utilized in the proposed approach. A deep convolutional autoencoder extracts images features and reconstructs images, and a fully connected deep time delay neural network learns the dynamics of a robot task process from the extracted image features and motion angle signals. The 'Nextage Open' humanoid robot is used as an experimental platform to evaluate the proposed model. The object folding task utilizing with 35 trained and 5 untrained sensory motor sequences for test. Testing the trained model with online generation demonstrates a 77.8% success rate for the object folding task.
AB - We propose a practical state-of-the-art method to develop a machine-learning-based humanoid robot that can work as a production line worker. The proposed approach provides an intuitive way to collect data and exhibits the following characteristics: task performing capability, task reiteration ability, generalizability, and easy applicability. The proposed approach utilizes a real-time user interface with a monitor and provides a first-person perspective using a head-mounted display. Through this interface, teleoperation is used for collecting task operating data, especially for tasks that are difficult to be applied with a conventional method. A two-phase deep learning model is also utilized in the proposed approach. A deep convolutional autoencoder extracts images features and reconstructs images, and a fully connected deep time delay neural network learns the dynamics of a robot task process from the extracted image features and motion angle signals. The 'Nextage Open' humanoid robot is used as an experimental platform to evaluate the proposed model. The object folding task utilizing with 35 trained and 5 untrained sensory motor sequences for test. Testing the trained model with online generation demonstrates a 77.8% success rate for the object folding task.
KW - Humanoid robots
KW - learning and adaptive systems
KW - motion control of manipulators
KW - neurorobotics
UR - http://www.scopus.com/inward/record.url?scp=85034700187&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85034700187&partnerID=8YFLogxK
U2 - 10.1109/LRA.2016.2633383
DO - 10.1109/LRA.2016.2633383
M3 - Article
AN - SCOPUS:85034700187
VL - 2
SP - 397
EP - 403
JO - IEEE Robotics and Automation Letters
JF - IEEE Robotics and Automation Letters
SN - 2377-3766
IS - 2
M1 - 7762066
ER -