<p>In many applications of an flexible joint robot (FJR), achieving high tracking performance is challenging due to various factors such as external disturbances, parametric uncertainties, nonlinearities, unmodeled dynamics, and the need of high-order state derivatives. This paper proposes a fixed-time extended state observer (FxTESO)-based adaptive fixed-time integral sliding mode controller (AFxTISMC) for the tracking trajectory of the FJR with different initial conditions (ICs), using the singular perturbation (SP) method. First, the SP approach is applied to simplify a high-order FJR into two reduced-order slow and fast subsystem to mitigate the cause of noise amplification. Second, two FxTESOs are integrated with both subsystems to estimate the unmeasurable velocities and the lumped disturbances, which enables designing the AFxTISMC for the fast system with a large bandwidth. Third, two fixed-time integral sliding mode controller (FxTISMC) is designed for the both FJR subsystems to properly address the ICs problem while achieving fixed-time convergence of link tracking errors even in the presence of issues mentioned above. Fourth, adaptive laws are utilized in forming the AFxTISMC the to reduce chattering and relax the requirement to know the upper bound of disturbances. Simulations and real-time experimental validation confirm the superiority of the proposed method in terms of strong robustness, high tracking accuracy, and fixed-time convergence of the tracking errors. These results prove that the proposed method significantly improves the tracking performance, reducing the RMS error by 67.92% to 98.33% compared to existing methods across multiple test cases.</p>