Towards supervised autonomous task completion using an in vivo surgical robot

Laparoscopy is a minimally invasive alternative to traditional abdominal surgery. Unlike traditional surgery, a laparoscopic procedure can be completed using small incisions. The use of these small incision results in reduced pain to the patient, shorter recovery times, and less trauma to skin, muscle and other tissues. However, these benefits to the patient are offset by the increased difficulty to the surgeon performing the procedure. These difficulties include reduced dexterity, reduced perception, and longer procedure times. The use of small in vivo robotic devices in minimally invasive surgery is one possible solution to these problems. The movement of these devices is not constrained by the position of the entry incision, because the devices would be completely intracorporeal. In addition to improving the quality of minimally invasive surgery, devices such as these could be used to perform supervised autonomous surgical tasks over a high latency communications channel. This dissertation discusses the contributions of the author towards the goal of creating surgical robots that can perform supervised autonomous surgical tasks. First, the design and testing of several in vivo robotic devices is described. Next, experimental results using visual quality metrics comparing in vivo cameras to laparoscopes are presented. Next, experiments conducted with the cooperation of NASA during the NEEMO 9 mission are discussed. These experiments compared the usefulness of in vivo robots to laparoscopes in simulated surgical tasks. Next, a sterilizable camera device was designed, and then tested in three survivable pig surgeries. The device was shown to cause no tissue damage or infection, and was used as the sole visual feedback device for a laparoscopic cholecystectomy. Finally, a prototype system was developed to demonstrate that a dexterous manipulator device could be used to perform supervised autonomous surgical tasks. A closed loop controller using visual feedback was implemented to control the robot. Benchtop tests demonstrating supervised autonomous task completion are presented. The author believes this work represents some work in using in vivo surgical robots to automate surgical tasks