Overview / Usage

In a surgery, a robotic arm can assist a doctor by handling equipment or holding things like clamps, scalpels, scissors, etc. One of the procedures the robot has to follow in order to fulfill this task is to detect where the object is inside the work-cell.
The project will focus on recognizing different surgical instruments inside a robotic surgery work-cell using four Kinect RGB-D sensors, a ceiling mounted operating theater robot and transfer learning with a pre-trained CNN. The kinect sensors will acquire information from different viewpoints. The data will be used to classify objects taking into consideration the scenery constraints such as the light of the room, the color of the instruments and the occlusion that may happen due to a person or the limb of the robot blocking a sensor.