To address the issues of missing evaluation standards for the usability of three-dimensional gesture interaction in spinal decompression surgical robots and excessive subjectivity, a usability evaluation system for medical robot gesture interaction was constructed through a fusion of support vector data description (SVDD) and fuzzy comprehensive evaluation (FCE) methods. Based on the characteristics of spinal decompression surgery and the constraints of the medical environment, a four-level index system covering learnability, efficiency, user satisfaction, and safety was established using the Delphi method. The SVDD algorithm was used to detect anomalies in the ranked data of 15 orthopedic experts, identifying abnormal experts and implementing adaptive deweighting through the minimum volume hypersphere model. A multi-level fuzzy comprehensive evaluation model was established, combining triangular fuzzy number membership functions and expert questionnaires to conduct a comprehensive evaluation of three spatial movement gesture schemes for robotic arms. The results show that the SVDD algorithm identified 3 abnormal experts, with the Kendall coefficient of consensus among experts significantly improving from 0.420 5 to 0.692 3, an improvement of 64.6%. Weight analysis indicated that interaction satisfaction and reliability were key indicators, with weights of 20.86% and 18.40%, respectively. Usability evaluation revealed that the thumb-pointing gesture received the highest score of 3.97, the overall motion control gesture scored 3.65, and the index-pointing control gesture scored 3.51. It can be seen that the SVDD-fuzzy comprehensive evaluation method significantly improves the objectivity and discriminability of medical robot usability evaluation. The results verify that the thumb-pointing gesture is the optimal interaction scheme for the laminectomy surgical robot.