Although direct phase current control of brushless DC motor prevents commutation current ripple at low speed, it occurs at high speed which has not received the deserved attention in the literature. Dealing with this current ripple is of practical significance because commutation becomes more influential for high speeds as its duration and the current ripple's amplitude increase with speed. This paper concerns with the successful application of a fuzzy logic estimator (FLE) as an expert control technique to minimize the so-called current ripple profitably. Phase current reference waveforms are programmed as a function of commanded current, angular position, and commutation angle which is adjusted online by the developed FLE as per the motor working condition. The presented approach renders the current references with changing but equal slopes during commutation to keep the other phase current constant at all times. A genetic algorithm is also deployed to optimize the FLE's rule table. Unlike the reported researches, this study does not require calculating commutation time, and use of torque observer and/or commutation detection circuits. The acceptability of our proposal is widely illustrated by simulated and experimental results using DSP TMS320F28335, which signifies that prolific performance toward commutation current ripple minimization is achieved.