In this study, we attempt to estimate the time-varying wrist impedance in the peg-in-hole task. We employ a wearable impulsive perturbation generator called the snap motor, which has advantages such as little disruption to voluntary movement during task execution and no need for a force (torque) sensor for impedance estimation. The task phase is divided into some subphases, and the wrist impedance at each subphase is estimated. In addition to the measurement data such as wrist motion and electromyogram, we ask the participants about their intention of the task execution. The experiment was carried out with three healthy participants, and their wrist impedance depending on the task phase was estimated. The obtained wrist impedance of each participant was consistent with their EMG pattern and intention of motion control respectively.