A new algorithm to maintain lateral stabilization during the running gait of a quadruped robot

• This is a new algorithm to keep the lateral stability of quadruped robots, which is a major problem in advanced walking robot in a running gait.
• The method can keep the locomotion parameters (stride frequency and stride amplitude) when a disturbance occurs during the running gait.
• It does not need an additional package to be included when a disturbance occurs, since the algorithm can manage these disturbances in a natural way.
• A fast approach is added that allows the algorithm to be executed at a frequency of 1 kHz (sample time=1 ms).

This paper presents a new uncoupled controller (based on a Kinetic Momentum Management Algorithm, KMMA) which allows a quadrupedal robot, whose operation is simple and fast, to run using a symmetrical gait patterns in a wide variety of scenarios. It consists of two tasks: calculating the lateral position and speed of the fore swinging leg when it next makes contact with the ground; and controlling the roll angle by mean of inertia forces using the stance legs.The KMMA provides the benefits of modulation and the synchronization typically presented in CPG (Central Pattern Generation) models. Furthermore, it is able to maintain the locomotion parameters (such as stroke frequency of gait pattern) when the robot runs in a highly disturbed environment, thus resulting in a lower energy consumption. Additionally, the uncoupled scheme of the leg makes the operation computationally cheap, thus avoiding the use of a Virtual Actuator Control or a Hybrid Zero Dynamics.The performance of the KMMA has been verified by means of co-simulation (using ADAMS and MATLAB) with a highly realistic model of a quadruped robot with uncoupled legs. The performance of the algorithm has been tested in different situations in which the following variables have been varied: frontal velocity, turning ratio, payload, external disturbances and terrain slope. Successful results in terms of stability, energy efficiency, and adaptability to a complex locomotion environment have been obtained.