Inertial Parameter Identification in Robotics: A Survey

This work aims at reviewing, analyzing and comparing a range of state-of-the-art approaches to inertial parameter identification in the context robotics. We introduce “BIRDy (Benchmark for Identification Robot Dynamics)”, an open-source Matlab toolbox, allowing systematic formal performance assessment considered algorithms on either simulated or real serial robot manipulators. Seventeen most widely used found scientific literature are implemented compared each other, namely: Inverse Dynamic Model with Ordinary, Weighted, Iteratively Reweighted Total Least-Squares (IDIM-OLS, -WLS, -IRLS, -TLS); Instrumental Variables method (IDIM-IV), Maximum Likelihood (ML) method; Direct approach (DIDIM); Closed-Loop Output Error (CLOE) Input (CLIE) Nonlinear Kalman Filtering (DDIM-NKF), Adaline Neural Network (AdaNN), Hopfield-Tank Recurrent (HTRNN) eventually set Physically Consistent (PC-) methods enforcement physicality using Semi-Definite Programming, namely PC-IDIM-OLS, PC-IDIM-IV, PC-DIDIM. BIRDy is robot-agnostic features complete pipeline, from generation symbolic kinematic dynamic models process itself. includes functionalities excitation trajectory computation as well collection pre-processing experiment data. In this work, proposed first evaluated simulation, following Monte Carlo scheme 6-DoF TX40 RV2SQ industrial manipulators, before being tested platforms. The robustness, precision, computational efficiency application different investigated discussed.