In this thesis, a study on the realization of intelligent controls is made in three fields: overall structure of intelligent machine, fault detection and tolerance, and intelligent fusion of sensor data. Firstly, a novel structure of the organization level of the hierarchical intelligent machine is proposed. The proposed model has the ability to search complete activities by learning without referring to the knowledge base of the incompatible event pairs. Also it has fault tolerant capability by using directly a performance measure from low levels as a probability. Application of the novel structure to a simple mobile robot organization problem is considered to demon-strate its effectiveness by computer simulations. Secondly, we propose fault detection and tolerance of the locomotion of the hexapod robot in even terrain. The fault stability margin is defined to represent potential stability a gait can have in case a sudden fault event occurs to one leg. And using this, the fault tolerant quadruped periodic gaits of the hexapod walking over perfectly even terrain is derived. It is demonstrated that the quadruped gait derived in this paper is the optimal gait the hexapod can have maintaining fault stability margin nonnegative. And the modified tripod gait is proposed for the hexapod to continue optimal locomotion after fault occurrence. Thirdly, a multisensor decision fusion strategy with fuzzy measure theory is proposed. This fusion strategy is constructed for finding optimal decisions through fusing decisions derived from a suite of parallel sensors. Each element of the possible target object set is assigned a-priori a fuzzy measure for all the sensors meaning the subjective weights of the sensors on each object. Through recursive sensing process all the possible target objects have the cumulative decision measures(CDM), which are derived from fuzzy measures to represent possibilities that one object is the target to be identified. The properties and applicability of the proposed algorithm is analyzed.