Trajectory tracking control of flexible continuum robots is challenging due to their inherent compliance and high nonlinearity. Many related works exclude the control of the end's orientation, i.e., only the end's position is considered. In this article, a differential-algebraic equations (DAEs) model-based instantaneous optimal control (IOC) framework for the end's position and orientation cooperative tracking of a cable-driven tensegrity continuum robot (TCR) is developed. Based on the tensegrity concept, a TCR is designed first as the control object, which can achieve multimode deformations such as bending, scoliosis, contraction, and the S- or J-shape. Then, the actuation of cables is introduced as the system kinematic constraints from the view of multibody dynamics so that a control-oriented model of the TCR can be built by DAEs. Subsequently, the original continuous trajectory tracking problem is approximated for a series of IOC problems at each discrete time slot. Finally, considering the constraints of control input saturation, a linear complementarity problem was derived for solving these IOC problems. The method provides an easy-to-implement and unified framework for addressing the trajectory tracking control issues of cable-driven continuum robots, which can improve the control performance of the position-only tracking controllers and exploit the TCR's advantages to handle more application scenarios. The advanced performance and potential applications of the proposed controller have been evaluated via several numerical simulations and experiments on the TCR prototype.

Deep learning networks can be applied to the field of intelligent prediction of part surface roughness. However, the surface roughness samples of parts have the problems of high collection cost, unbalanced categories, and complicated data distribution, which inevitably limit the application of deep learning network models in the field of intelligent prediction of part surface roughness. To solve these problems, this article proposes a novel data augmentation method based on CoralGAN for prediction of part surface roughness, which introduces the domain adaptive method deep coral function to help optimize the network parameters of the generator of generative adversarial network (GAN). Specifically, the vibration signal collected during processing is converted into frequency spectrum data and input into CoralGAN. The training of the generator is guided by coral loss, that is, the distance between the covariances of the real samples and generated samples features, not just the statistical consistency of the traditional GAN. Experiments have been carried out on the three-axis vertical machining center. Research shows that the proposed method can improve the prediction accuracy of part surface roughness to 95.5%.

To understand the thermal behavior of the spindle, the radial thermal drift errors (RTDEs) of the spindle are tested in this paper. To analyze the influence of environmental temperature to the RTDEs, the RTDEs are tested at 0 r/min. Afterward, the RTDEs of the spindle and the temperature values of key points are tested at different speeds. The thermal deformation process of the spindle is analyzed and verified by thermal imaging testing. The RTDE models of ten thermal deformations are established. Also, the relationship between compensation value and tool length, determination criterion of thermal deformations, algorithm flow, and parameter identification process are presented by using an inverse model. The compensation effects are verified using both simulation and experimental data. The results indicate that high accuracy can be achieved for all rotating speeds, with parameters identified using data of one speed. The strong robustness of the proposed model is proved.