DB_navi8

본문

Twisted-String Actuators(TSAs) 

 

This research aims to develop high performance robot joints. To improve the performance of the robot joint, we proposed a robot joint using Twisted-String Actuators(TSAs) and proposed a variable radius pulley that can improve robot performance by controlling the speed ratio of the drive. It is confirmed that the performance of the robot joint is improved through various experiments.

 

3a0ed6ceef33ff33fb87c4f56f074258_1509885634_4444.jpg

Fig. 1. Title


3a0ed6ceef33ff33fb87c4f56f074258_1509885634_4906.jpg 

Fig. 2. Title 

 

 

<Reference>

[1] Park, Jihyuk, et al. "Design and control of antagonistic robot joint with Twisted String Actuators." Ubiquitous Robots and Ambient Intelligence (URAI), 2016 13th International Conference on. IEEE, 2016.

[2] Park, Jihyuk, et al. "Note: Position/torque control of antagonistic robot joint with high-compliant twisted string actuators (TSAs)." Review of Scientific Instruments 87.12 (2016): 126107.

[3] Park, Jihyuk, et al. “A passively adaptive variable radius pulley for tendon-driven robot” Under review.



Title


In this project, we design electric manipulator and implement a valve motion planner. It’s a robot to be put in a situation when power plant accident occurs. To close the valves inside the plant, tens of Nm of torque is required. We develop hardware platform and algorithm architecture for opening these valves.


3a0ed6ceef33ff33fb87c4f56f074258_1509885687_5564.jpg
Fig. 3. Title


Title

The capture point(CP) is the point on the floor onto which the robot has to step to come to a complete rest. The desired zero moment point(ZMP) can be produced by the CP dynamics. Using the model predictive control(MPC), ZMP reference is used to calculate the desired position and velocity of center of mass(COM) which are translated to the joint angles using the inverse kinematics. When the CP is perturbed, The controller tries to apply appropriate ZMP to shift the CP to the desired CP position. That makes online step position adjustment possible.   Object detection using the convolutional neural network and depth map generation are used to generate the online walking pattern.  For autonomous walking, robot have to detect the obstacle and calculate the depth from the object. The final goal of this research is autonomous walking based on the robust control and vision system.

3a0ed6ceef33ff33fb87c4f56f074258_1509885780_2057.jpg

Fig. 4. title


3a0ed6ceef33ff33fb87c4f56f074258_1509897830_4726.png

Fig. 5. title


Title

This research is about dynamics and automatic control for humanoid robot. This is the following research of disaster response robot TRUST, which was previously developed for executing missions in a nuclearized site. This robot more focuses on dynamics and automatic control. The purpose of the robot is riding an alpine ski. Alpine ski is very fast and dynamic sport. To ski on a real ski slope, there need some skills not only balancing itself, but also detecting the flags and avoiding them. Although it may not be a hard work to human, but very challenging problem to robot. We are preparing this robot to ski, PyeongChang Winter Olympics 2018.

3a0ed6ceef33ff33fb87c4f56f074258_1509885836_3473.jpg
Fig. 6. Title






;
MSC Lab.4F(#3434~#3446) ID B/D(N25), School of Mechanical Aerospace & System
Engineering : Division of Mechanical Engineering, KAIST, 373-1, Guseong-dong, Yuseong-gu, Daejoen, Korea, 305-701.
Tel:+82-42-350-3268 l Fax:+82-42-350-5201