In medicine, robots can be applied as a part of complex and computer-assisted systems for diagnosis, preoperative planning, surgery, post-operative patient care, and hospital logistics. Surgical robot systems can improve the existing operative procedures in terms of better efficiency, accuracy, and greater reliability of performance. Since the operating target in neurosurgery is not visible, the use of robots requires spatial patient registration. The spatial patient registration is an alignment of patient images acquired by means of an appropriate kind of scan technology with a patient located in the operating room (OR). Registration, in general, is a fundamental problem which occurs in many scientific fields, such as machine vision, image processing, robotics, and medicine, and it denotes transformation of two data sets into one coordinate system. The research proposed in this doctoral thesis addresses major elements of spatial patient registration in robotic neurosurgery: localization of the patient in the medical images and in the OR, rigid point-based registration, and automation of the overall patient registration procedure. This implies a good knowledge of the state-of-the-art methods in robotic surgery, the development and implementation of new methods and algorithms, and measurements that evaluate the achieved results. In order to improve the image space localization, a novel algorithm was developed; it uses a unique approach combining machine vision algorithms, biomedical image filtration methods, and mathematical estimation methods to determine the centre of each individual fiducial marker. A novel correspondence algorithm and a framework for an automatic patient registration procedure using freely distributed fiducial markers in the application of a robot in neurosurgery were established. Both the image space and the physical space localization, and, subsequently, the registration, are executed autonomously and do not require the additional employment of the medical personnel. For localization in the physical space, a concept of robot localization strategy was introduced, implemented, and tested. Localization strategies use specific approach angles, orientations and types of movement of a robot during the fiducial marker localization procedure in the physical space and positioning to the target points. Influence of the robot localization strategy on the overall application error of a robot system used in frameless stereotactic neurosurgery was measured and analysed.