Abstract: Human-robot collaboration (HRC) assembly serves as a practical carrier for human-centric smart manufacturing of Industry 5.0. Compliant control is crucial for ensuring efficient and safe HRC assembly. However, traditional position-based control is unsuitable for HRC assembly dominated by motion-following control, presenting challenges such as difficulty in identifying HRC intent and end-effector oscillation. A joint variable impedance compliant control for collaborative robots based on the particle swarm optimization (PSO) was proposed. Firstly, the overall framework was constructed. Subsequently, the dynamic model of the robot was established, and its dynamic parameters were identified. The PSO algorithm was then introduced into the adaptive variable impedance model to optimize the performance of the collaborative robot in both force tracking and force maintenance tasks. Finally, simulations were conducted in Matlab/Simulink, and the effectiveness of the proposed method was validated through experiments including zero-force drag control, constant force step control, and variable force following control.