Soichiro Hayakawa

Analysis and synthesis of driving behavior based on mode segmentation

This paper presents the development of the modeling of human driving behavior based on an expression as a hybrid system (HS) focusing on the driverpsilas vehicle following task. In our modeling, a relationship between the diverpsilas sensory information and the output of driver is expressed by the piecewise ARX (PWARX) model, which is a class of the HS. As the sensory input, the range between vehicles, range rate, and time derivative of the area of the back of the preceding vehicle (called KdB) are considered. Also, the pedal operation is considered as the output. The identification problem for the PWARX model is solved using the clustering based technique. By introducing the PWARX model, it becomes possible to find not only parameters appearing in the operation in each mode but also parameters in the logical switching (decision making) conditions among them from the measured driving data. Furthermore, the obtained model is exploited for the design of assist system especially focusing on the switching condition of the ON/OFF of assist. The usefulness of the proposed assist system is confirmed by experiments.

Modeling of Human Behavior in Man-Machine Cooperative System Based on Hybrid System Framework

Recently, the demand for a man-machine cooperative system, where the machine assists the human operator, is rapidly growing in the industrial fields. To meet this demand, the human model is required to design the suitable assist controller in the man-machine cooperative system. This paper presents a new human behavior model based on a piece-wise affine model which is a class of hybrid dynamical system, and apply it to a sliding task. Since the human behavior is considered to consist of several primitive motions expressed by continuous dynamics and a decision-making expressed by the discrete switch, it seems to be natural to introduce the hybrid system modeling. Particularly, the decision strategy for the number of discrete modes is addressed by using a hierarchical clustering technique, and the measured data are classified into several modes. Then, each primitive motion in each mode is identified based on the affine model. Finally, the switching conditions among modes are identified by applying support vector machine to the classified data. The obtained piece-wise affine model can quantitatively represent both primitive motions and decision-making in the human behavior

Trajectory Control of Manipulators Based on Discrete Time Learning Control with 2-Delay Input Control Technique

A learning control based on repetitive operations of robotic manipulators is one of the most promising method to realize high speed and high precision control for robotic manipulators. The joint servo system usually has an unstable zero in discrete-time domain. This fact leads to unstable learning, when a sampling period is small. In previous paper, we have proposed a new learning algorithm based on so-called 2-delay input method, By using the 2-delay input method we can stabilize the unstable zero and realize the stable learning even if the sampling period is small. However, concrete algorithm for parameter design and experimental results have not been shown yet. In this paper, we propose new design algorithm of two parameters in 2- delay input system especially to suppress ripple of the trajectory and show some experimental results to verify effectiveness of the proposed method.

2A2-M11 Measurement and Analysis of EEG in Driving Situation on Simulator

Recently, the analysis and application of biosignal is attracting great attention. In particular, the estimation of human internal state based on the electroencephalogram (EEG) signal is a promising strategy because of its huge amount of information. In this paper, first of all, the EEG signal under the driving situation is captured synchronizing with the behavioral signals such as the pedal operation. Then, the ‘mode’ underlying the observed EEG signal is extracted by applying the stochastic switched autoregressive (SS-AR) model. Finally, the meaning of each mode is investigated from viewpoint of the correspondence with the driver’s internal state.

Shaft insertion for moving object by using robot manipulator with one dimensional PSDs sensor

The shaft insertion task of the industrial robot is very important and frequently performed in assembly operations. We have studied the shaft inserting robot for the moving object by using a PSD (position sensitive detector). We propose a new position error detecting sensor in which four 1 dimensional PSDs are arranged in cross around the shaft. This sensor can detect the position error between the center of shaft and the hole and the inclined angle in real time. The sensor can detect the position error during the shaft inserting operation. At first, we show the detecting principle of the position and the incline angle of this sensor. Next, we show the configuration of the robot system with this sensor. Using this sensor system, the robot can insert a shaft into no inclination hole of the moving object at a velocity of 80 mm/s with a clearance of 50 /spl mu/m without chamfering. From the experimental result for the inclined angle detection, the characteristic of the inclined angle detection differed from the one of the theoretical output. We verify the difference between the experimental result and the theoretical result.

Shaft insertion into moving object using robot manipulator with cross typed PSD

Shaft insertion is a frequent and important operation in automatic parts assembly. To insert a shaft, the active method using the force sensor and the passive method using RCC (remote center compliance) system have been studied. We develop the sensor using a cross typed PSD (position sensitive device) as the position sensor and the shaft inserting system for the moving object with the robot system applying this new sensor. The robot with this sensor can insert a shaft into the hole of the moving object with 100/spl mu/m of the clearance at a speed of 40mm/s. Our proposed system is effective for the shaft inserting into the hole of the moving object.

Experimental Study on Acquisition of Optimal Action for Autonomous Mobile Robot to Avoid Moving Multiobstacles

The principal aim of this study is to show how an autonomous mobile robot can acquire the optimal action to avoid moving multiobstacles through the interaction with the real world. In this paper, we propose a new architecture using the hierarchical fuzzy rules, fuzzy performance evaluation system and learning automata. By using our proposed method, the robot acquires the fine behavior to move to the goal, avoiding moving obstacles, simultaneously by using the steering and velocity control inputs. Also we show the experimental results to confirm the feasibility of our method.

Discrete-time learning control for robotic manipulators based on 2-delay input method

A learning control based on repetitive operations of robotic manipulators is one of the most promising methods to realize high speed and high precision control for robotic manipulators. For implementation the learning algorithm should be represented in the discrete-time domain. When the sampling period is small, the joint servo system will have an unstable zero. This leads to unstable learning. In this paper, the authors propose a new learning algorithm based on so-called 2-delay input method. By using the 2-delay input method, the authors can stabilize the unstable zero, and realize the stable learning even if the sampling period is small. Also, the authors show some simulation results to verify the effectiveness of their proposed method.<<ETX>>