This presentation deals with the paradigm of Central Pattern Generator (CPG) and with the study and modification of a well-known model in the field of quadruped robot locomotion. A CPG is a neural network of the Central Neural System (CNS) that generates periodic motor commands for rhythmic actions without sensory feedback. The CPG model for quadruped locomotion that we will discuss about in this presentation is the one described in the article listed as a reference (Golubitsky, Buono). The presentation is divided into three parts. In the first part, the main ideas underlying the concept of CPG and gait theory will be briefly outlined. In the second part, we will describe the main steps of the design of the application. In this part, the structure of the CPG will be described, which is a eight-neuron neural network whose purpose is to simulate the primary gaits in a quadruped. Then, we will describe the internal dynamics of each neuron, the complete dynamics of the system with its main mathematics properties. After this, we will expose the original contribution of our work . We modified the original model in order to simulate primary gait shifts during the operation. To achieve such a goal, we endowed the model with few modifications deriving from control theory. In the third part, the main steps of the implementation process will be described. At first, we implemented the original CPG and reported a simulation example. Then, we implemented the modified CPG and reported a simulation example.
Central Pattern Generation (CPG), Periodic Solutions, Gait Theory, Gait Simulations
Pietro-Luciano Buono · Martin Golubitsky. (2001)
Models of central pattern generators for quadruped locomotion, I. Primary gaits,
Mathematical Biology, 42(4):291 - 326.