Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
Convolutional Neural Network for Parameter Identification of a Robot
Ist Teil von
Advances in Data-Driven Computing and Intelligent Systems, 2023, Vol.653, p.523-534
Ort / Verlag
Singapore: Springer Singapore Pte. Limited
Erscheinungsjahr
2023
Link zum Volltext
Quelle
Alma/SFX Local Collection
Beschreibungen/Notizen
Parametric identification is a crucial issue due to the possibility of emulating systems for motion control, collision detection, manufacturing, and other scientific areas. However, the conventional methodologies need an optimized trajectory to find all the parameters quickly. This paper shows an identification method based on a convolutional neural network to extract the dynamic parameters of a cartesian robot. First, the variables of position, velocity, acceleration, and torque with a set of parameters create an image with a conversion method. Second, the backpropagation algorithm trains the convolutional neural network to return the robot’s parameters. Finally, a time-spectral evaluation distance gives the affinity of the results. The simulations show that the identification achieved 0.9916, and the experimental robot met 0.9196.