An identical parallel machine scheduling problem with earliness and tardiness penalties is studied. First, a hybrid method adopting the meta-heuristics, e.g., the simulated annealing method combined with list scheduling techniques is proposed, where the heuristic procedure MPA (minimum penalty increase assignment) and the SD (start-time decision) are newly introduced. Then, through several computational experiments, the effectiveness and the potential of the proposed method are investigated.

This paper introduces a coevolutionary approach to genetic algorithms (GAs) for exploring not only within a part of the solution space defined by the genotype-phenotype map but also the map itself. In canonical GAs with the fixed map, how a large area of the solution space can be covered by possible genomes and consequently how better solutions can be found by a GA rely on how well the genotype-phenotype map is designed, but it is difficult for designers of the algorithms to design the map without a-priori knowledge of the solution space. In the proposed algorithm, the genotype-phenotype map is improved adaptively during the searching process for solution candidates. It is applied to 3-bit deceptive problems as a kind of typical combinatorial optimization problem, which are well-known in that the difficulty against GAs can be controlled by the genotype-phenotype map, and shows fairly good performance beside a conventional GA.

We consider an extended class of flexible job shop scheduling problems. First, we translate the problem into a mathematical programming formula, i.e., a mixed-integer programming problem. This makes it possible to apply standard packages of mixed integer programming solvers and, while lots of computational time is required in general, to obtain the optimal schedule. Then, in order to seek the schedules close to the optimal for larger-scale problems, we newly design a solution method by adopting genetic algorithms based on the formula. Through some computational experiments, the effectiveness and the possibility of the proposed approach are examined.

We apply the reinforcement learning to acquire a gait pattern of a quadruped locomotion robot. The advantage of the reinforcement learning for such a problem is that no exact robot model is needed for calculating the prescribed teaching signals, but one simply needs to evaluate the results of trials and generates the reinforcement signals. As a result, the robot can acquire by itself a walking pattern suitable to its structure, dynamics and environments. We use here a tightly coupled modular actor-critic structure with stochastic gradient ascent. The computer simulations show that it could generate various stable walking pattern suitable to the environment and dynamics of the robot. We also apply the proposed method to an experimental real robot and deal with the learning process for getting the walking pattern.

A neuro oscillator is applied for the generation of biped locomotion. Dynamics of each neuron is described by a set of nonlinear differential equation, and the coupled system of four neurons is used. Two problems were solved in this study. The first problem is the stability analysis as nonlinear oscillator. It is shown that the system generates periodic solutions for appropriate set of free parameters in the nonlinear differential equations. The stability chart in the parameter plane, depicting the stable, unstable and walkable region, was obtained with various motion patterns corresponding to several parameter sets. Next problem is a design aspect of the walking pattern. The harmonic balance method is successfully applied for this problem. It enables us to determine the free parameters in the oscillator's equations to satisfy a prescribed walking mode. The methods proposed here provide a theoretical issue for analysis and design problems encountered in the nonlinear neuro oscillators.

We present the theory and practice of permutation coding as a new tool for very low-bit-rate image compression. Conventional source coding deals with the data information of signals, while the permutation coding achieves compression through efficiently representing the positional information (i.e., position permutation) caused by ordering the data information into order statistics. A set of four theorems is presented. The first one reveals the information-theoretic relationship between data and permutation information and the rest solves the efficient coding problem. For this, novel tools from finite group theory are applied to derive a compact form of representation for permutation, called permutation-cyclic-representation (PCR) vectors, with which various regularities and constraints in the structure of positional information are displayed, whereby the coding is made very easy using a runlength and Huffman method. A block DCT-based permutation coding algorithm (the BCPC) is developed attempting to combine the DCT's excellent features of energy packing and magnitude ordering that are found to be amenable to permutation coding. This mutually beneficial characteristic significantly reduces the coding bit-rate. Simulation results are provided for real images, showing an improvement by 3-4 dB in the peak-SNR index as compared to those representing the state-of-the-art.

A method of compliant and collision avoidance control is presented. The collision avoidance motion is first generated in off-line routine and its data are stored as as a sequence of desired joint angles. Next, virtual impedance model-following control is employed as force feedback control. Redundancy is used for minimization of the variance from the desired joint angle during compliant motion. In particular, the computation time for the inverse kinematics at each sampling time was reduced by using approximate values of Jacobian matrix. The stability of the control system is analyzed and the effectiveness of the proposed method is discussed.>