readme.txt

==================================================================
Description

- The video material is provided by the authors of the paper:

	"Compliant Terrain Adaptation for Biped Humanoids without Measuring Ground Surface and Contact Forces" 
	by Sang-Ho Hyon

	submitted to IEEE Trans. on Robotics.

- The material is downloadable from http://ieeexplore.ieee.org
- The file size is 12.5 MB
- The file format is MPEG4
- We recommend the latest version of QuickTime Player to play the video.
- The file includes four video clips which show the experimental results using our biped humanoid robot. See the details below.

--------------------------------------
Contents

1. Adaptation to unknown external forces (Section II)

This video demonstrates the robot can keep the balance under unknown external disturbances. The balancing is achieved by a full-body gravity compensation combined with a simple feedback control of CoM.
The video also shows the robot can interact to external force applied arbitrary contact points. The time evolution of the CoM and CoP, as well as the ground reaction forces, is provided in the paper, where the actual (measured) trajectories are tracking to the desired ones with small errors.

2. Adaptation to uneven ground surface (Section IV)

This video clip shows that a humanoid robot is balancing on (unknown) uneven ground. A wooden block is put under the right foot, and then suddenly removed or put again by an operator. The operator is sometimes pushing the robot. In this experiment, the impedance controller is deactivated to allow the foot move freely when it does not contact with the ground. The robot keeps self balance and compliantly changing its stance, showing the adaptability to rough terrain. The time evolution of state variables is shown in the paper.

3. Adaptation to inclination of low friction (Section IV)

This video clip shows that a humanoid robot is balancing on a wooden seesaw. The incline of the seesaw is manually changed at random by the human operator. The robot keeps self balance by measuring the center of mass and applying the recovery contact forces while compliantly changing its posture, showing the adaptability to rough terrain. In this example, the impedance control for preventing the feet from a slip because the friction of the wooden plate is not enough. The time evolution of state variables is provided in the paper.

4. Impedance control test (Section IV)

This video shows the effectiveness of the impedance control between the feet. We put hydraulic oil on the floor (a human could not walk on this surface). Initially, the robot was standing at an initial posture with high position gains. In this case, the robot was not balancing, but one could easily move the robot on the ground because there was little friction between the feet and the ground. Then, the robot smoothly switched to force control mode, and compliantly balanced. If we applied the impedance control in addition to the nominal balancing control, the robot could keep its stance while balancing. Next, we removed the impedance controller. Then, the robot slipped as soon as a small disturbance is applied.

--------------------------------------
Contact information

If you have problems in playing the materials, please contact to:

	Sang-Ho Hyon (sangho@atr.jp)

	ATR, Computational Neuroscience Laboratories
	2-2-2 Hikaridai, Kyoto 619-0288, Japan

==================================================================