Physically feasible dynamic model identification and constrained control of robotic arms: A case study on the ViperX-300 6-DoF robotic manipulator

Abstract

The demand for novel robotic applications continues to grow, driven by the need for advanced solutions to complex tasks. These tasks often require the incorporation of dynamic models into the development of control schemes. Consequently, model identification has become a critical step in creating accurate dynamic models for robotic arms. However, the identification process can yield parameters that lack physical significance, resulting in feasibility issues that lead to unrealistic and potentially unstable dynamic models. This challenge is particularly pronounced in small robotic arms, where the dynamics are highly sensitive to parameter variations. To address these challenges, this article presents a systematic and comprehensive approach to the model identification of robotic arms while strictly enforcing physical feasibility. The proposed approach is validated on the ViperX-300 6DoF robotic manipulator, which lacks a pre-existing dynamic model, making it an ideal candidate for testing our method. To further validate the proposed model identification approach and assess the suitability of the obtained model for the ViperX-300 6DoF robotic manipulator in safety-critical applications, we design an explicit reference governor, which is a model-based constrained control strategy. Experimental results show that the identified model achieves sufficient accuracy for safety-critical applications, showcasing the effectiveness of the proposed model identification approach and the reliability of the identified model for the ViperX-300 6DoF robotic manipulator.