59 fromData vs Model in Medical Robotics Workshop - IROS 2023 Computer Vision News of surgical expertise like the United States, medical errors are the third cause of death, right after cancer and heart diseases and well before stroke and car accidents, causing around 250,000 deaths every year. Approximately, 40% of these errors happen in the operating room and 50% of the resulting complications are avoidable. The mission of my research is to address these problems by building systems and interfaces for human skill augmentation in robot-assisted surgery (RAS). My focus has been on two instances of the skill augmentation spectrum: (i) Surgical skill acquisition, where I design interfaces to improve the motor skill learning aspect of surgery, and (ii) Task execution, where I design autonomous systems to perform repetitive tasks, such as suturing, so that surgeons can focus on the more demanding ones. The key insight of my research is that leveraging robot’s unique capabilities, that are different from humans, can revolutionize human skill augmentation in RAS. This is unlike the related work in this area, that has the implicit assumption that RAS basically replicates open surgery, through smaller incisions, ignoring many of the robot’s unique capabilities. For instance, existing RAS platforms only have one camera, imitating the single pair of eyes of the main human surgeon in open surgery. Moreover, autonomous systems in RAS perform tasks following the same steps as humans. In contrast, in my research, I identify some of the robot’s unique capabilities and design systems and interfaces to leverage these capabilities to better augment humans in RAS. One example of these unique capabilities is that robots can have more than one pair of eyes, unlike humans. To leverage this capability, we designed and tested a stereoscopic 6-DoF (Degrees of Freedom) surgical camera to provide an additional view of the surgical scene, in addition to the existing camera in the RAS platform. Moreover, we developed a visual-motor alignment method that enables the surgeon to easily control the surgical tools with respect to the view of the additional camera. We also developed a multi-camera, multi-view system, where the user can see two views of the surgical scene simultaneously, as a picture-in-picture view. One of the views comes from the original surgical camera and the other comes from our additional one. We showed that such system improves surgical training and skill assessment compared with the traditional single-view system. Another example is that robots can have adjustable “interpupillary distance”, unlike humans. To leverage this, our additional stereoscopic camera design is “side-firing”. That is, the two vision sensors are placed on the side of the camera body. This is unlike the current “end-firing” surgical
RkJQdWJsaXNoZXIy NTc3NzU=