Paper Title
Human Pose Estimation based Robust Gait Generation and Estimation in Humanoid Robot

This paper proposes a methodology and observations to achieve adaptive and autonomous gait generation and correction mechanisms in humanoid robots. The methodology used in this paper has been implemented in Robotis Darwin Humanoid Robots for dynamic gait generation in uneven terrain and inclined plane. The paper follows a two-thread process where the first thread deals with generating the gait pattern and after implementing the generated gait, the second thread has a feedback mechanism for gait correction. 2D Pose Estimation is used on Human Gait videos to extract gait cycle pattern which is normalized and applied to the Robotic actuators. An IMU sensor present in the Centre of the Robot provides the feedback data for the gait correction. The cumulative score of the angles generated by the Gait Generation Unit and the Feedback Unit are fed to the actuators to perform a robust walk cycle on the Humanoid Robot. Keywords - Robotics, Legged Locomotion, Robot Dynamics, Robust Control