A safe ex-vivo application of an autonomous medical robot with AI

Abstract

While most robot platforms in clinical settings are still controlled by a human operator, synergy of robot technology and artificial intelligence will eventually enable fully autonomous surgical robots. Here we show such a combined robot system where neural networks ensure safety for the patient while increasing the accuracy of the system. The system is able to autonomously remove hard material and was tested by drilling a pre-operatively planned access cavity in a predetermined tooth in a phantom head. The performance of the system was validated by a deviation analysis of 56 teeth prepared by the robot system with a Micro CT scanner. The safety level of the system and the response time to potentially hazardous situations was also assessed. The results show that the system is capable of drilling openings in a consistent manner (trueness: 0.34mm, precision: 0.2mm). The neural networks allow the system to detect external forces and visual occlusion and show excellent detection of hazardous situations). The performance of the system surpasses human performance and meets clinical standards.