This paper introduced a coordinated framework for network-based humanoids. The humanoid robot consists of three sub-systems: the walking/mobile platform, and the perception system. The sub-system should be scheduled according to the task sequence. The task planner scheduled the actions for the diverse sub-systems, and the XML script had been presented for describing tasks. The design of a real-time distributed robot control system based on IEEE-1394 for the humanoid robot was presented. Based on the whole-body control framework, the humanoid robot is able to walk and balance its body. It is also able to plan its path, taking into consideration the location of an obstacle, and move itself by biped walking or wheeling. The location of an object was estimated with the particle filter; thus, real-time object tracking was possible. To track the face which was detected and recognized by AdaBoost and PCA methods, a mean-shift algorithm using bilateral filtering was used according to its robustness to illumination changes. Task operation by a humanoid robot was successfully performed in an experimental situation. Humanoids which are made to function similarly to humans are a futuristic technology that can reduce the work load in work and home environments. This is a core technology of human welfare, and humanoids must be developed to be human-friendly, efficient, and similar to humans, so that they may co-exist with humans in a common environment. The ultimate aim of the humanoid research is to develop a humanoid which can provide services in real home environments utilizing a network