Deep Learning Aided Dynamic Parameter Identification of 6-DOF Robot Manipulators

Abstract

Generally, structural uncertainty of the robot dynamics system refers to model error caused by parameter identification, unstructured uncertainty is the unmodeled dynamic characteristic. No matter how elaborate modeling methods are used, there always be uncertainty. Therefore, this paper applies deep learning for the first time to aid robot dynamic parameter identification of 6 degrees of freedom robot manipulator for compensation of uncertain factors. Firstly, the relatively accurate prediction of torque is obtained by the physical dynamic model using the parameters identification method (errors are less than 10% of the maximum torques). Secondly, we propose a novel deep neural network based approach called Uncertainty Compensation Model (UCM) to compensate the torque error introduced by the uncertainty. The UCM mainly composed by proposed Input Control Module (ICM) and Error Learning Models (ELMs) based on Long-Short-Term Memory and attention mechanism. The proposed ICM, which effectively avoids the unnecessary interference, is used to control valid input for ELMs. The ELMs, consisted by ELM units, concern extracting salient features from sequence data to predict the joint error. Also, this paper summarizes the effects of valid input, timestep and attention mechanism on the performance of the UCM. Finally, the verification of parameter identification and torques compensation is carried out by a Universal Robot 5 manipulator. Compared with the prediction torques of physical dynamic model, the proposed UCM has a good ability to capture the friction characteristics and compensate for the error of local maximum torques, which effectively solves the deficiency of the physical dynamics model and improves prediction accuracy (errors are less than 6% of the maximum torques).

Red neuronal artificial para la extracción de parámetros dinámicos de robots a partir de información incompleta de su movimiento

“…Sus resultados de más de 100 segundos muestran que el helicóptero se estabiliza en la posición deseada. Wang et al (2020) utilizan una red neuronal long short-term memory (LSTM) en conjunto con el algoritmo de mínimos cuadrados para la identificación de parámetros dinámicos de un robot, sin embargo, se necesita de la optimización de una trayectoria de movimiento para que el robot identifique parámetros dinámicos. Liu et al (2020) utilizan una red neuronal artificial en conjunto con el modelo dinámico de un robot para estimar el torque.…”

Design of motor cable artificial muscle (MC-AM) with tendon sheath–pulley system (TSPS) for musculoskeletal robot

“…In order to fulfill the demands of various high-precision force sensing environments, such as surgical machines, Su et al [30] introduced two multi-layer neural network approaches to enhance the sensing accuracy of the end-of-manipulator tool. Wang et al [31] applied deep learning approach called uncertainty compensation model for the first time to aid robot dynamic parameter identification of six DOF robot manipulator for compensation of uncertain factors. Akhmetzyanov et al [32] proposed the application of deep learning methods for kinematic error compensation of four-DOF cable-driven parallel robot.…”

Introducing convex BiLSTM-based controller applied to 3-PSP spatial parallel robot manipulator

“…However, the community warmed up to LSTM’s usage. Wang et al (2020) used a deep-learning approach to aid a 6-degrees-of-freedom robot manipulator dynamic parameter identification, where the approach compensates for the torque error using LSTM, and the attention mechanism avoids unnecessary interference. Yu (2017) proposed a convex-based LSTM network for fast learning purposes for identification, which shows good results on the speed requirement of system identification.…”