Robotic machining has obtained growing attention recently because of the low cost, high flexibility and large workspace of industrial robots (IRs). Multiple degrees of freedom of IRs improve the dexterity of machining while causing the problem of redundancy. Meanwhile, the performance of IRs, such as their stiffness and dexterity, is affected by their position and posture obviously. Therefore, a redundant posture optimization method for robotic milling is proposed to improve the machining performance of the robot. The multiple characteristics of the robot are considered, including the joint-limit, singularity and stiffness, which have symmetry in its workspace. Firstly, the joint-limit is regarded as the constraint. And a symmetrical and effective constraint method is proposed to simply guarantee that all the interpolation points can avoid joint interference. Then, the performance indices of singularity and stiffness are designed as the optimization target. On this basis, the piecewise-global-optimization-strategy (PGOS) is proposed for redundant optimization. Owning to the PGOS, all the given planned tool points in their corresponding segment are considered simultaneously to avoid the gradual deterioration in traditional methods, which is especially suitable for the machining process with a continuous path. Moreover, the computational load of the optimization solution is considered and limited by the designed segmentation strategy. Finally, a series of comparative simulations are conducted to validate the good performance of the proposed method.

Accurate dynamic model and response prediction of industrial robots (IRs) are prerequisites for production optimization before actual operation. In this study, a hybrid-driven dynamic position prediction (HDPP) approach integrating a parametric dynamic model (PDM) and learning-based residual error compensators (RECs) is developed to estimate the actual position of a robot end-effector based on the reference input trajectory. Firstly, a PDM consisting of a flexible dynamic model of the mechanical system and a servo system model is constructed as the primary predictor in HDPP. Meanwhile, a reinforcement learning (RL)-based parameter identification method is presented to obtain independent dynamic parameters, which integrates a CAD model, least squares estimation, and RL. Then, an REC based on the temporal convolutional network long short-term memory (TCN-LSTM) is proposed for each joint to compensate for the residual error after PDM prediction. A TCN is employed as the input of LSTM to extract and compress the discontinuous features, which can enhance the compensator’s accuracy and stability. Additionally, a dynamics-integrated (DI) dataset construction scheme is developed for network training to boost the prediction accuracy. Finally, a series of experiments and comparative analysis are preformed to validate the performance of HDPP in terms of prediction accuracy and stability.

Top-hat-type harmonic drives are widely used in industrial robots and high-end equipment due to their large hollow structure, high load capacity, and high precision. However, unlike cup-type harmonic drives, the transmission error characteristics of top-hat-type structures are influenced by assembly-induced inclination, while the underlying error propagation mechanism remains insufficiently understood. To address this research gap, this paper reveals that the periodic transmission error of top-hat-type harmonic drives is closely associated with assembly-induced inclination and eccentricity rather than isolated machining deviations. Assembly-induced inclination periodically alters the meshing state between the flexspline and circular spline, thereby generating characteristic harmonic components in the output signal. Based on this error propagation mechanism, separate transmission error models are established for the flexspline, circular spline, and wave generator, and an integrated prediction model specifically tailored for top-hat-type harmonic drives is further developed. Transmission error experiments are conducted using an integrated joint test platform, and model parameters are identified through frequency-domain feature matching. Experimental results demonstrate that the proposed model accurately captures the dominant transmission error components. Compared with the conventional fitting-based model, the maximum absolute residual and mean square error are reduced by approximately 65% and 95%, respectively, indicating improved prediction accuracy. The proposed method provides a theoretical basis for transmission accuracy analysis and precision compensation of top-hat-type harmonic drives.