Variable Motor Stiffness In Nao Humanoid Robot Stability

I also worked in a project in the humanoid robot soccer team, Cerberus of Bogazici University in Turkey with Cetin and Tekin Mericli. The platform we worked with was the Nao humanoid robot. In my project, I investigated methods to enhance the stability of the bipedal walking movement. To achieve this goal, the key variable I manipulated was the stiffness value of the motors in the leg joints of the robot. The concept of stiffness of a motor captures the idea of how resistant the motor is to external forces. In other words, if we position the joint at a certain angle and try to exert force on it to change this angle externally, the joint will demonstrate resistance directly proportional to its stiffness value.

The key idea was that if the robot can loosen its leg joints as it puts more weight on them, it would have a less rigid and more stable walking motion. To realize this approach, I wrote a feedback control system which controls the stiffness values in the leg joints depending on stance of the robot. To evaluate the stance of the robot, a number of sources of input, such as gyrometer, electric current values in the leg joints and foot pressure sensors at the feet were used.

In the initial experiments, we showed that this feedback mechanism leads to a more stable walking gait. In the graph below, we demonstrate how the gyrometer values in the y-axis, the axis perpendicular to the chest of the Nao, that read the stability of the robot (red) in that axis is significantly smaller with the feedback controller (blue).

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License