Humanoid Crawling
[Improving the Crawling of a Humanoid Robot]
This semester project aims at improving the crawling of a humanoid robot based on contact feedback. I did experiments in the step down environment which acts on the forelimbs or the hips.
At first, I used central pattern generators (CPGs) which are neural networks located in the spine of vertebrates with the rhythmic patterns controlling locomotion. And also I took dynamical systems theory which allowed me to build general design methodologies for such controllers. For dynamical system in iCub, I optimized parameters with optimization algorithm PSO. Secondly, I integrated sensory feedback in the system and simulated four different experiments. Finally, I compared the results and drew conclusion on the project.
I compared four experiments in terms of a force in left and right hand and trajectory of robot
-
Without Feedback
-
With Feedback, but only feedback to joint angles of elbows
-
With Feedback, fixed feedback to joint angles of both elbows and hips
-
With Feedback, adaptive feedback to joint angles of both elbows and hips
Documents
Midterm_Defense_JunghyunKim.pdf
Videos
iCubCrawling_StepDown_wo_feedback.mp4
iCubCrawling_StepDown_with_fb_elbow.mp4