QDSEGA is one of efficient learning algorithm for hyper-redundant systems. However, to accomplish the learning process, it required many iterations of trials. One of the most significant cause of the problem is lack of efficiency of learning, which is observed at the beginning of the learning.In this paper, we consider the mechanism of the lack of efficiency and solve the problem, we propose a method that the state space is devided into several classes based on physical dynamics. We improve QDSEGA by embedding the proposed dividing method into the QDSEGA, and demonstrate the effectiveness of the improved QDSEGA by applying it to a locomotion task of real snake robot.
This paper proposes a novel adaptive control approach for a robot manipulator to track a desired force signal while interacting with its unknown dynamic environment. Based on the bio-mimetic studies of human motor control functions, we propose a 2 degree of freedom adaptive control approach. In this approach, both force feedback and feedforward controllers are involved in the robot's control system. The feedback control is set as constant while the feedforward controller is adjusted online adaptively to approach the inverse of the force control transfer function. Using this approach, exact force tracking without any loop delay can be realized. The effectiveness and problems of this control approach are studied through computer simulations and robot experiments.
In the petrochemical industry, we have been required to optimize a production schedule for the efficient management. Determining how to provide naphtha to the petrochemical plant is one of the most crucial problems in the optimization of such production scheduling. Typically, purchased naphtha, which is provided from the sea berth, is first stored temporarily in storage tanks, and after that, is provided to the plant, according to a predetermined production schedule. Then we are required to find pairwise disjoint routes between the sea berth and a tank, between two distinct tanks, or between a tank and the plant. In this paper, we formulate the problem of finding routes in a tank network as a problem of computing disjoint paths in a graph, and propose an algorithm for enumerating all sets of disjoint paths based on the graph theory. Our algorithm first constructs a tree-shaped data structure for each pair of prescribed vertices which represents all paths between the pair of vertices, and then computes disjoint paths efficiently by traversing the tree-shaped data structures. We also evaluate the practical performance of our algorithm by conducting a computational experiment based on the case in Showa Denko K.K.
This paper is concerned with stability analysis of parallel-flow heat exchanger equations with boundary feedback. The parallel-flow heat exchanger equation is super-stable when the boundary inputs are identically zero. However, when a pole assignment problem is considered, a suitable output feedback law needs to be added to the equation in order to produce finite poles. In this paper, it is shown that a C0-semigroup generated by the closed-loop operator, which consists of the parallel-flow heat exchanger equation and an output feedback law, satisfies the spectrum determined growth assumption.
This paper is concerned with discontinuous output feedback stabilization of a class of nonholonomic systems in a port-controlled Hamiltonian form. First, in order to obtain a dynamic feedback, an integrator is added to the system via a generalized canonical transformation. Second, we clarify an equivalence between asymptotic stability of a state feedback system and that of the corresponding output feedback system. An output feedback stabilization method derived based on this equivalence. Furthermore, some numerical examples show the effectiveness of our technique.