Tesis "CPG-based control for adaptive legged robot locomotion using visual information: a hardware implementation approach"
Alumno: José Hugo Barrón Zambrano
Asesor: Dr. César Torres Huitzil
Sinodales: Dr. Arturo Díaz Pérez, Dr. José Gabriel Ramírez Torres, Dr. Marco Aurelio Nuño Maganda, Dr. Héctor Hugo Avilés Arriaga
The design of locomotion control systems for legged robots is a challenge that has been partially solved, however, there are still design and technological problems to be tackled. Legged locomotion consists in determining the best sequence (locomotion pattern) for lifting off and placing the feet over the floor. Several proposed solutions taking inspiration from locomotion mechanisms found in animals have been explored, being one of the main reasons the fact that they provide clues for the design of efficient legged robot locomotion systems. The aim of this thesis is to design and implement an adaptive locomotion control for legged robots based on the concept of central pattern generators (CPG).
The main focus of this work is on adaptive goal-directed behaviors by integrating basic visual perception information to modulate CPG locomotion according to a sensed situation. Although many works have aimed to develop robust legged locomotion controllers, relatively few works have focused on adapting the theory and technology to fully practical embedded implementations. In this contribution, a reconfigurable digital CPG controller implementation able to generate several gaits for quadruped and hexapod robots was developed. To adapt the locomotion patterns, legged robots are usually equipped with different kinds of sensors. The information of these sensors must be processed and classified in order to modulate the robot locomotion.
However, this task is not easy due to it is necessary to find a relationship between the sensed information and the locomotion modulation. Hence, a mechanism based on visual perception to modulate the locomotion patterns was proposed. The integration mechanism was built from a combination of Finite State Machine (FSM) and fuzzy logic. The research results show that the CPG controller and its implementations are able to generate different locomotion gaits and these gaits can be modified through changes in their parameters or using feedback signals. Experimental results confirm that the CPG-based locomotion implementation can effectively drive a real robot to perform rhythmic locomotion.
The results of the integration mechanism show that is feasible the generation of complex behaviors from the modulation of simple periodic signals generated by CPG structures through the integration of visual information.