In robotic surgery, the surgeon controls the surgical robot through human-machine interfaces. The most common interfaces are for hands so that the surgeon can only control at most two robotic instruments at one time while three or more instruments may be needed at the same time in some procedures, e.g., robotic endoscopic/laparoscopic surgery. Current practices to these cases, i.e., having one assistant control the additional instrument, is not ideal because fine collaboration between the surgeon and the assistant is required, and delayed actions and even safety issues due to communication errors may occur. These limitations can be avoided if the surgeon could control all the tools on his/her own simultaneously. This thesis work developed a novel foot-controlled human-machine interface, enabling the surgeon to control the additional robotic surgical tool using the foot, together with the two tools that are controlled by the hands. This foot-controlled interface controls a robotic surgical instrument in continuous direction and speed based on the motion of the foot in four degrees of freedom (DoFs), i.e., foot left/right translation, forward/backward translation, lateral/medial rotation, and toe-up/down rotation. The foot force/position signals are collected through eight feedback-sensing modules of the interface surrounding the foot. Each feedback-sensing module consists of a spring and a load cell. The foot's position can be determined by the deformation of the springs which is calculated based on the forces measured by the load cells. The whole foot operation is separated into elastic and isometric modes. In the elastic mode, the springs are not fully compressed and thus the foot can move freely with reaction forces from the springs (i.e., passive force feedback); in the isometric mode, the foot's motion is limited by the boundary (some springs are fully compressed) but its force is not limited and thus can still be used for output commands. The seamless transition between the two modes provides the user with rich proprioceptive information and meanwhile enlarges input ranges that are limited only by his/her strength. The interface also features a singularityfree workspace with a neutral central home position. The weight of the foot is well supported, and the friction of the interface is minimized to limit operation fatigue.i An analytical model based on the kinematics and statics of the foot interface was derived for output command calculation; however, this model does not consider the distinct motion behaviors of different users. Therefore, a data-driven mapping approach using independent component analysis was proposed to adapt the interface to the specific user's motion pattern, which effectively reduces the deviation between users.Three studies were conducted on the developed interface: (1) Controlling of an industrial robotic arm for path-tracking and button-pressing tasks and comparison to a foot button interface. The proposed interface was about 30% faster and 60% smoother than a conventional foot butto...