A new type of multiple-variable fuzzy logic controller is proposed, called the two-layer multiple-variable fuzzy logic controller. In the first layer, various fuzzy decision-making logics are applied to fuzzy variables. The feedback is the weighted sum of the fuzzy variables obtained from fuzzy inference. In the second layer, the weight of each variable is determined by fuzzy inference and system parameters. The two-layer fuzzy control algorithms are applied to a nonlinear, cart-pole system. Computer simulations show the effectiveness and better performances of the two-layer fuzzy logic controllers.>

A fault-tolerant parallel processing method based on the linear arithmetic code is proposed for computation of robot kinematics. This method can detect and correct online a single error in the resultant matrix or vector with low hardware and time redundancy. The method also can be extended to multiple errors in computation. It can be applied to various hardware architectures for robot kinematics computation.>

The implementation of a nonlinear control algorithm in multiprocessor systems is addressed. A single bus structure and a multiport structure are proposed for real-time computation. The computing task is decomposed into a reasonable number of subtasks by considering both the execution and the communication time of subtasks. In order to deal with the real-time performance, the models of multiprocessor systems with single-bus and multiport-memory architectures are established using Markovian birth-death queuing models. The intercommunication time can thus be estimated using analytical formulas. On the basis of the estimated intercommunication time and task execution time, the necessary number of processors in the multiprocessor system for a specific real-time robot control algorithm can be estimated.>

A novel neural network model, based on the gradient system theory, is introduced. The proposed design approach solves the problem of parasitic limit points. This could have significant impact on many potential applications, particularly in the area of pattern classification/recognition. The design approach, the development of the Lyapunov function, the stability analysis, and the convergence characteristics of the neural network are discussed in detail. Design examples and simulation results are presented to illustrate the design process and the convergence characteristics of the proposed neural network. One example shows its application in pattern recognition.>

Digital computer simulations and computer graphic approach have been successfully applied to research in robotic systems. In order to design various kinds of control algorithms of robots in real time environments, real time simulations and real time dynamics in computer graphics need to be developed. This paper deals with the computation complexity and the approaches to reduce the computation time for real time simulations of robotic systems. Possible parallelism and pipeline processing in the simulation algorithm are investigated. Different algorithms of simulations are compared. Using today's multiprocessor systems and VLSI technology, real time simulation of robotic systems is possible.