1M. Sc. [Engg.] Student, Department of Electronics and Electrical Engineering, M. S. Ramaiah School of Advanced Studies, Bangalore-560 058
2Professor and Head, Department of Electronics and Electrical Engineering, M. S. Ramaiah School of Advanced Studies, Bangalore-560 058
3Assistant Professor, Department of Electronics and Electrical Engineering, M. S. Ramaiah School of Advanced Studies, Bangalore-560 058
Online published on 18 February, 2020.
This study concerns the development, modelling and control of a bipedal humanoid robot and algorithm to generate dynamic walking pattern using sensory feedback. The implementation of only servo control for robots with walking systems is not sufficient. Even having stable motion pattern and well tuned joint control, a humanoid robot can fall down while walking. Therefore, an adaptively generating gait pattern method is needed, which allows the robot to walk stably in various environments, such as on rough terrain, up and down slopes. To achieve this, the robot has to adapt to the ground conditions with the foot motion and maintain its stability with a torso motion. The gait synthesis is based on the analysis of human's gait pattern. The proposed algorithm is applied to generate the natural and stable gait pattern of the biped robot.
A small-scale humanoid robot prototype was constructed and used as a test platform for implementing the concepts described; the humanoid robot has totally 18 DoF (Degree of Freedom). Each arm has 3 DoF, each leg has 6 DoF with 2 DoF in ankle allows to control step size and frequency. The main controller communicates with sensor unit and servo motors in real time.
The gait trajectory algorithm is composed of two kinds of trajectory. The first one is Inverse Kinematics (IK). The second one is feedback to the controller from Inertial Measurement Unit (IMU) and the servo motors in the robot, which senses the robot's velocity, Centre of Gravity (CoG) and posture and it allows to generate and modify motion commands in real time. The motion commands generated comprises of 4 walk states like left leg rise, left leg down, right leg rise and right leg down each with 7 sequences for all the 18 servo actuators to obtain 1 walk cycle with 2 steps, the number of the walk cycles can be specified by the user. The robot was able to walk with a dynamically-stable gait either passively down a ramp, or actively on a flat or slightly uphill surface. From this study we observed that the robot has successfully functioned as driven by the MATLAB code.
