Fuzzy Inference based Operation Training Framework with Application to Microvascular Anastomosis

Most conventional training schemes require trainees to perform repetitive operations. In this way, novices usually require a long learning period and lack personalized training assistance. To address the above issues, we propose a new training framework based on fuzzy inference. Firstly, a modified fuzzy C-means (FCM) classifier is utilised to partition the tasks based on human motor behaviours. Operation performance is comprehensively assessed in each subtask by task-based criterion and the corresponding results are fed back to the trainee in real time during the subtask execution. Once the trainee reaches the operational standard of each subtask, he/she can proceed the next one. Distinguished from traditional repetitive training without intervention, the trainee can leverage purposeful modifications and repetitions in a closed-loop manner. The expertise degree of trainees can lead to differences in training time, avoiding unnecessarily long training sessions for experienced trainees. The proposed method is therefore suitable for the parallel training with different levels of operation. Comparative experiments on microvascular anastomosis task have demonstrated higher training efficiency of the proposed training strategy.