Collision Localization Algorithm for Surgical Robots Fusing Image and Force Data

Collision localization is an important function of force feedback; it can improve the safety and dexterity of surgical robots. In this work, a novel collision localization algorithm is provided to detect the collision position of bone blocks or surgery instruments during a robot-assisted surgery. The novel algorithm is based on the fusion data of pre-operative images and intra-operative force. Compared with methods that are only based on vision or force sensors, the novel algorithm no longer depends on the geometrical information of the collision surface. Instead, it requires the collision position from the three-dimensional image data predicted by the pre-operative plan, using real-time force data. Additionally, a simulative orthognathic surgery was adopted to verify the effectiveness and accuracy of the fusion algorithm. Eight positions on the skull model were chosen as the test collision position. For each of the test point, the experiment was repeated 10 times. The total mean error of the experiments is 1.18 +/- 0.431 mm. The results demonstrate that the novel algorithm proposed in this work can realize collision detection, and its accuracy and reliability is superior to the previous Force-Torque-sensor-based collision position detection algorithms. Using the novel algorithm, the surgical robot can locate the obstacle and replan the path in time during the operation.