Intelligent Kinematic Control of a Quadrupedal Mobile Robot

This work presents a control method for quadrupedal locomotion utilizing periodic gaits, alongside a neural network-based solution for the inverse kinematics problem inherent to these robots. The proposed method evaluates the stability and execution speed of both periodic-continuous and periodic-discontinuous gaits. Based on this analysis, the optimal gait for the robot locomotion can be chosen. For the inverse kinematics problem of the 12-DOF quadrupedal robot, a feed-forward neural network-based solver was proposed. Different training methods were analyzed. The performed simulation experiments demonstrate the effectiveness of the proposed computation and control approach.