In this thesis, an active part-mating algorithm using multisensor is presented and implemented. The algorithm assumes unknown environment and takes advantage of multisensor data fusion. The part-mating is performed in two stages. In the first stage, a hole with unknown x, y positions is searched using a vision sensor. The hole position thus found has an error of 2-3 min. In the second stage, a 6 axes force sensor and a vision sensor are utilized to insert a peg into the hole. The data from these two sensors are fused using the generalized Bayesian method of the Maximum Likelihood Estimation(MLE) to decide the angle along which the robot moves the peg. The sensors were modeled by repeating an experiment to acquire the statistics of measurement data on the angle. It was shown that the established sensor model could be expressed as a normal distribution. The result shows that the utilization of multisensor fusion method provides some advantages on part-mating in unknown environment.