This paper investigates the position control of induction motor in the presence of model uncertainties, load torque disturbances, unknown input gains, and unmeasured rotor velocity. A model-free position tracking controller is proposed based on novel memorybased data-driven extended state observers and dynamic surface control design, under which the induction motor can track the reference position without any priori knowledge of model parameters. In the observation loop, memory-based data-driven extended state observers are developed to estimate the model uncertainties, load disturbances, as well as the control input gains without the measurement of rotor velocity. The convergence of estimation errors is guaranteed without persistently exciting condition, and the true value of control input gains are reconstructed in finite-time. In the control loop, a dynamic surface controller is developed based on the proposed observers, such that the model-free position tracking control of induction motors can be achieved. The observer and controller are designed in modularity and form a cascade system, which is proved to be input-to-state stable. Simulations illustrate the effectiveness of the proposed model-free position tracking controller for induction motors.