It is important to estimate the road model and vehicle state for Lane Departure Warning Systems and Lane Keeping Systems. One of the conventional methods is to estimate these parameters using the Kalman filter. Recently, the H∞ filter has been derived based on H∞-theory. The H∞ filter is formulated as the robust state-estimation filter to the modeling errors and disturbances (the initial state and noises). In this paper, a estimation method of the road model and vehicle state using H∞ filter is proposed. In addition, we develop a simple method to check an existence condition of the H∞ filter per time step and a fast algorithm in case that the size of the observation vector is larger than the size of the state vector. The efficiency of the proposed methods is demonstrated by simulations.
Aiming at an excellent schedule with respect to due-date related criteria, we propose a method that allows a large lot-size job to be effectively pre-empted and split by a smaller job. In order to determine the best pre-emption strategy, it is necessary to review the jobs and the possible timings for pre-emption. The proposed method involves two basic procedures within a general search procedure. The first procedure gives an initial configuration schedule where large lot-size jobs are assigned by the GA method so as to maximize due-date earliness, and where small lot-size jobs are appended to the large-lot size job assignments. The second procedure is a schedule generated by a forward simulation where every large lot-size is split into 'unit jobs' and small lot-size jobs can mix among the unit jobs. At the end of the procedure, the unit jobs are rejoined to create a larger job. The search procedure introduces three parameters to determine the jobs and the timings to pre-empt and split the larger job. The parameter-space-search-improvement (PSSI) method is used to efficiently search for an excellent schedule. The performance of the proposed method is investigated using a simple flow-shop model by comparing results obtained by different methods. The characteristics of the job pre-emption are also discussed by assuming different scheduling conditions such as release times and due-dates.
Recently, efficient spatial data acquisition and visualization have been receiving more attention from the view point of digital city, VR museum including digital archives. Generally, in order to perform object modeling through digital images, line or feature extraction and stereo matching are performed, and many matching methods such as area based matching, feature based matching have been proposed. In particular, line gives important information for object extraction, and satisfied 3D results depend on rigorous line extraction and matching. With this motive, the authors have developed robust line matching which is performed by line extraction and line tracking. The line extraction was performed by Canny operator, and line matching was performed using optical flow and trifocal tensor. In order to apply the line matching method for modeling of historical object, 3D modeling for the “Koma house” through image sequences was investigated in this paper. The Koma house was built in 17th century, and designated as national important cultural assets in 1971.
We discuss a force control problem for a constrained one-link flexible manipulator based on distributed parameter model. The proposed control law consists of bending moment at the root of the flexible manipulator and its derivative, which regulates simultaneously the force and the rotational angle of the motor without the angle and the force information. And it ensures the asymptotic stability of the closed-loop system. As the control law is derived based on the distributed parameter model, we can avoid the drawbacks resulting from finite dimensional approximation.
This paper presents a new change detection method with the aid of subspace identification. The proposed method is based on monitoring a change in variance of a statistic generated by a recursive subspace identification algorithm. An asymptotic property of the statistic is presented. Without changes during sampling, it is shown that, under relevant assumptions, the statistic converges in probability to a stack of noise vectors multiplied from the left side by a Toeplitz matrix. A numerical example illustrates that the proposed method can detect changes in the dynamics of a system, without being disturbed by changes in the spectral density function of an input signal which is not our concern.