In this paper, a speed sensorless induction motor drive is introduced which is direct vector controlled in a universal field-oriented (UFO) reference frame. This chosen reference frame is easily linked with direct and indirect rotor, stator and air gap field orientation control schemes of the induction machine (IM) drives using a stator to rotor virtual turn ratio. Based on partial feedback linearization theory, a nominal speed controller is designed first, regardless of the system parameter uncertainties and external load torque disturbance. Then, in the perturbed condition, a so-called totally invariant sliding mode (SM) controller is employed that has low SM chattering with no reaching phase. SM controller ensures the sliding motion through the entire drive system state trajectory. In addition, a fuzzy-adaptive algorithm is used to estimate the optimum upper bound of the motor lumped uncertainty function due to the rotor moment of inertia, friction cofficient and external load torque disturbance. As a result, the proposed Controller combines the merits of fuzzy inference mechanism and the adaptive algorithm. Furthermore, a model reference adaptive system (MRAS) is derived to estimate the rotor speed and a simple method is also proposed for on-line estimation of the stator resistance. Finally, the effectiveness of the proposed speed controller scheme is verified by computer simulation.