Abstract: Aiming at the time-optimal operation problem of machine tool loading and unloading manipulators under kinematic constraints, an improved grey wolf optimizer combined with the simulated annealing algorithm is proposed. Firstly, an angle-smoothing guidance mechanism is introduced to make the time-optimal trajectory satisfy the acceleration constraints. Secondly, a distance-adaptive convergence factor is designed to dynamically balance global exploration and local exploitation according to the population convergence state. Thirdly, an eight-direction partition guidance strategy in joint space is proposed to improve the search efficiency. Finally, a simulated annealing local search mechanism is embedded to enhance the global optimization ability. Simulation and experimental results show that the joint displacement, velocity and acceleration curves planned by the proposed algorithm are continuous, smooth and impact-free. The total running time is reduced by 2.99 s with a time compression ratio of about 33.22%. The planned trajectory meets kinematic constraints and practical operation requirements, which can provide a reliable scheme for time-optimal trajectory planning of industrial robots.