Human Posture Recognition Based on Neural Network in Robot Controlling
Published in NTCU Software Engineering Laboratory, 2013
Recommended citation: Cheng-Yan Guo., Liang Hung., Advisor: Chorng-Shiuh Koong, Ph.D. (2013). "Human Posture Recognition Based on Neural Network in Robot Controlling." Undergraduate Research.
http://GCY.github.io/res/undergraduate-research/paper.pdf
We study the possibility to control the robot using a natural user interface(NUI) and human posture recognition based on a neural network. Our humanoid robot platform is BeRobot, BeRobot is 16 degree of freedom(DOF) humanoid robot, we use the Microsoft Kinect received skeleton, then calculate the vector dot product of keypoint, input vector dot product to our posture recognition neural network(Fully Connected), output pre-define human posture, finally, transmit the posture to the robot through Bluetooth to control.
Keyword:Signal processing, Image processing, Robotics, Humanoid robot control, Nature user interface, Microsoft Kinect
Recommended citation: Cheng-Yan Guo, Liang Hung., Advisor: Chorng-Shiuh Koong. (2013). “Human Posture Recognition Based on Neural Network in Robot Controlling.” Undergraduate Research.