To address the complex challenge of identifying the contact state between a shaft and a hole and to improve the efficiency of robotic shaft-hole assembly tasks, a robotic shaft-hole assembly method based on variable admittance control is proposed. In this method, admittance control serves as the foundational force controller for shaft-hole assembly. On this basis, the Twin Delayed Deep Deterministic policy gradient (TD3) algorithm from deep reinforcement learning is utilized to optimize the parameters of the admittance controller. Additionally, a nonlinear reward function is designed, which not only prevents the assembly strategy from converging to local optima but also further accelerates the training speed of the assembly task. Experiments conducted with a collaborative robotic arm performing 15° inclined hole assembly demonstrated that the assembly efficiency of the variable admittance method was 9.6% higher than that of the fixed admittance parameter method, validating the feasibility and effectiveness of the proposed variable admittance shaft-hole assembly method.

Electric ducted fans (EDFs) have emerged as promising propulsion systems due to their compact design, high thrust density, and enhanced operational safety. Accurate prediction of aerodynamic thrust is essential for EDF design and performance evaluation; however, existing numerical studies have not yet provided a systematic comparison of the thrust-prediction capability of different k−ω-based turbulence models in micro-EDF applications. In this study, a dedicated thrust-measurement platform was developed for a 120 mm EDF, and experimental thrust data were obtained under three representative hover operating conditions. Based on these measurements, six turbulence models, including SST, SK$\omega$, BSL, GEKO, EARSM, and SST-$\gamma$(alg.), were evaluated using three-dimensional CFD simulations. The numerical model was assessed through thrust validation, centerline velocity comparison, power-consistency analysis, grid independence verification, and qualitative flow-field interpretation. A two-factor full-factorial analysis was further conducted to quantify the effects of rotational speed and turbulence model on prediction accuracy and computational cost. The results show that the turbulence model has a stronger influence on the normalized thrust-prediction error than the rotational speed factor over the investigated operating range. The SST-$\gamma$(alg.) model achieves the highest thrust-prediction accuracy, with an average relative deviation of 0.47%, but requires the highest computational cost. In comparison, the SST model provides a favorable balance between accuracy and efficiency, with an average relative deviation of 1.79% and an average computation time of 184.33 min, approximately 33% lower than that of the SST-$\gamma$(alg.) model. The centerline velocity and power-consistency results further support the comparative model assessment. Overall, this study provides an experimentally validated comparative reference for turbulence model selection in simulations of similar 120 mm EDF under hover conditions. Considering both prediction accuracy and computational efficiency, the SST model can serve as a practical turbulence model choice for engineering parameter optimization of similar micro-EDF configurations.