Knee arthroscopy is a common elective surgery that has been found to sometimes cause unintended damage to patients due to the limitations of the instruments, and is ergonomically challenging for surgeons. It has also been found that for surgeons to achieve baseline competency they must perform a high number of such surgeries. In order to reduce these limitations, a surgical continuum robot has been introduced, however, a clear understanding of the tasks that the robot will perform and the interface that surgeons would use to control the robot have not been identified. This research implemented a three-phased human-centred research through design approach, to explore how an orthopaedic surgeon and a robot can collaboratively conduct knee arthroscopic surgery together. In phase 1 an observational study and card sorting activity were conducted to understand the current limitations and challenges that orthopaedic surgeons face whilst performing knee arthroscopy. Phase 2 identified desired characteristics for each of the instruments, which can then be translated onto a robot. It established the desired roles that an orthopaedic surgeon and robot will undertake to perform knee arthroscopy through the use of a video-based card sorting activity using video data collected in the first phase. Phase 3 involved an interaction workshop with orthopaedic surgeons, theatre nurses, and roboticists using three different interfaces to control a surgical continuum robot, a robot that can continuously curve to produce movement like a snake, and identified the key qualities for the interaction between a surgeon and robot. The research outcomes have contributed towards better understanding the challenges that orthopaedic surgeons face when conducting knee arthroscopy. This research has also identified the desired relationship that should exist between an orthopaedic surgeon and a robot, which is that of `Artist-Instrument', with the surgeon being the artist and the robot an instrument. The roles that both the orthopaedic surgeon and robot will carry out have also been determined, with the surgeon maintaining high level decision making and the robot becoming a `clever instrument', providing greater manoeuvrability and stability. These outcomes produced design recommendations for the development of an interface to control a continuum robot to perform knee arthroscopy, providing natural mapping between the surgeon's movements and the movement of the robot that is comfortable to handle whilst also being a tangible interface to control a robot. Surgeons also expect to fully control the robot and be able to map the controls of the interface to suit their own preferences. The robot should also provide feedback to the user in the form of stereoscopic vision to enhance depth perspective, and haptic feedback to enable safe navigation around the joint. Augmented reality is also desired, thus allowing surgeons to perform the surgery over the patient in a natural pose. In addition, this research provided a unique approach to conducting a rapid in situ co-design workshop with healthcare professionals using a video-based card sorting activity that also included non-surgeons to examine surgery to identify limitations and discuss technology that could reduce these limitations. This research has the potential to impact the way in which knee arthroscopy is conducted to reduce an orthopaedic surgeon's physical and cognitive efforts and allow for better patient outcomes.