In the process of controlling underwater unmanned vehicles to accomplish target tasks, the operator’s implicit intentions are generally more difficult to recognize compared to explicit intentions which are easily captured and comprehended by machines. In this paper, we propose an implicit intention recognition method based on effective eye movement fixation. Using the Identification-Dispersion Threshold (I-DT) algorithm, we obtain effective eye movement fixation sequences. Based on these sequences, we apply the Support Vector Machine (SVM) algorithm to achieve classification of implicit intentions. Experimental results show that the proposed method achieves implicit intention recognition accuracy rate of 90.8%, and its performance surpasses traditional methods by 11.3%. The proposed approach facilitates the timely recognition of operators’ implicit intentions by machines and provides convenience for explicit interaction by means such as altering the interface layout, which is conducive to ensuring the smooth progression of human-machine interaction (HMI).