Robots and robot controllers are becoming more sophisticated. Consequently, the demands on the controller's operating system are increasing. The lower levels of robot control systems (indeed, most real-time control systems) are characterized by servo loops. This thesis examines servo loops and how they affect data communications within robot control systems. In the two systems described in this thesis the special characteristics of servo loops are exploited to enhance the data communications. H scIC is an operating system for hierarchies of servo loops. It uses rate monotonic scheduling for the periodic servo loop processes. H sc IC events (or processes) which are used to implement servo loops are not allowed to block. They will only surrender the processor upon completion or when preempted by a higher priority process. A non-blocking communication structure, Periodic Data Buffers (PDB's) was developed for inter-process communication. H scIC has been implemented and is used successfully in a controller for the Utah/MIT hand. G scANGLIA is a proposed real-time communication network. It is intended to allow the processors in a robot controller to be distributed within the robot. Thus the processors can be close to the sensors and actuators they control. Much of the traffic on such a network would be periodic. G scANGLIA uses a central controller which allocates access to the network. For the periodic traffic a fixed schedule, produced off-line, is used. For the aperiodic traffic round-robin polling is used. Unlike most protocols, messages do not contain the address of the destination node. Instead, the messages are labeled with the name of its contents. Each node examines each message and decides whether or not it is interested in the message. A special communication controller in each node (the Communication Memory Management Unit) examines and selects the messages. The result of this protocol is a network-wide common memory. In this thesis, the G scANGLIA protocol is described in detail and some preliminary analysis of its effectiveness in some real robot systems is given.