Michiya Matsushima

A learning approach to robotic table tennis

We propose a method of controlling a table tennis robot so as to return the incoming ball to a desired point on the table with specified flight duration. The proposed method consists of the following three input-output maps implemented by means of locally weighted regression: 1) a map for predicting the impact time of the ball hit by the paddle and the ball position and velocity at that moment according to input vectors describing the state of the incoming ball; 2) a map representing a change in ball velocities before and after impact; and 3) a map giving the relation between the ball velocity just after impact and the landing point and time of the returned ball. We also propose a feed-forward control scheme based on iterative learning control to accurately achieve the stroke movement of the paddle as determined by using these maps. Experimental results including rallies with a human opponent are also reported to demonstrate the effectiveness of our approach.

Realization of the table tennis task based on virtual targets

This paper describes how a table tennis robot with a flat paddle coordinates its movement in order to achieve efficient strokes for any given ball. We propose a method of generating stroke movement based on virtual targets that means the point at which the ball should be struck and the paddle velocity just before hitting the ball. These targets are predicted using inputs-outputs maps implemented efficiently by means of a k-d tree (k dimensional tree). The paddle approaches these targets by using a visual feedback control scheme similar to the mirror law proposed by Koditschek. The results of the implementation are also given to show the effectiveness of the proposed method.

Learning to the robot table tennis task-ball control & rally with a human

We propose a method of controlling the paddle so as to return the ball to a desired point on the table with a specified flight duration. The proposed method consists of the following three input-output maps implemented by means of Locally Weighted Regression (LWR): (1) A map for predicting the impact time of the ball hit by the paddle and the ball position and velocity at that moment according to input vectors describing the state of the incoming ball, (2) A map representing a change in ball velocities before and after impact and (3) A map giving the relation between the ball velocity just after impact and the landing point and time of the returned ball. We also propose a novel control scheme based on iterative learning control to accurately achieve the stroke movement of the paddle as determined by using these maps.

Visual Inspection of Soldering Joints by Neural Network with Multi-angle View and Principal Component Analysis

With the development of microelectronics technology, the demands of the automatic inspection system are ever increasing. The current trends toward miniaturization of components, denser packing of boards, surface mounting technology, and highly automated assembly equipment make the task of inspecting these defects more critical and more difficult for humans. In this paper, we achieved 0% of misjudgment by implementing training category intermediate samples. We also achieved cutting down the processing time and an increase of correct judgments by the improvement of inputs using principle component analysis and multi-angle image.

Estimation of current path area during small scale resistance spot welding of bulk metallic glass to stainless steel

Abstract The current path area is a significant factor in estimating the temperature distribution via numerical modelling for resistance spot welding. This paper presents a method for the estimation of the current path area at the faying surface during small scale resistance spot welding between bulk metallic glass and stainless steel. Observation of cross-sections and fracture surfaces reveals the welding process at the faying surface for both dissimilar and similar welding. The equipotential surface that depends on the difference between the contact area of the electrode-to-sheet and sheet-to-sheet interfaces is estimated by numerical modelling. The current path area at the faying surface is estimated by measuring the electric potential between the sheets, taking into account the current distribution.

Numerical Analysis of Self-Organizing Interconnection Process by 3 Dimensional Flow Dynamics

The growing importance of high integration on electronics demands novel interconnection methods replacing high-cost solder bumping or less reliable conductive adhesives. Self-organizing interconnection process using resin containing solder fillers has a possibility to achieve high-density joints satisfying both needs. Numerical study visualized the process and revealed that surface tension of molten fillers and resin viscosity determine the speed of conductive path formation.

Effects of Material Property and Structural Design on the Stress Reduction of the Joints in Electronics Devices

Electronics devices consist of silicon chips, copper leads, substrates and other parts which are jointed to each other with solder, conductive adhesive or other materials. Each coefficient of thermal expansion is different and it causes strain concentrations and cracks. We analytically investigated the stress reduction structure at the edge of the joints such as Sn-Ag-Cu solder or Cu/Sn alloy between the silicon chip and copper lead. At first, we examined the influence of the joint thickness and fillet at the joint edge on the stress. In the joint without fillet, the stress at the end of the joint increased depending on the thickness of the joint. The fillet of the joint increased the stress of the Cu/Sn alloy joint and the stress was increased depending on the thickness, though the fillet decreased the stress of the solder joint. We suggested the copper lead with slits to reduce the force of constraint. We compared the effects of the structure parameters of the slits on the stress reduction. The height was a more effective parameter than the width and the pitch. In the case of solder joint, the slits of the copper lead reduce the stress more effective in the thick joint than the thin joint. However, in the case of Cu3Sn joints, the slits reduced the stress more effectively in the thin joint than thick joint.

Microstructural development at weld interface between Zr-based glassy alloy and stainless steel by resistance microwelding

Zr-based bulk metallic glasses are expected to be welded to conventional structural alloys. Dissimilar welding of metallic glasses to stainless steel was carried out by resistance microwelding. The metallurgical analysis of the weld interface revealed the welding mechanism. A thin reaction layer was formed between the two liquid materials. The melting of stainless steel should be limited to obtain sound joints.

Movement of solder fillers because of the unevenness of interfacial tension in self-organization assembly process

Soldering or anisotropic conductive adhesives (ACAs) are used for assembling of devices with high-density area-array terminals. However, solder bump processes are high-cost and interconnects of ACAs are less-reliable. The purpose of this study is to develop a novel method. Self-organization assembly method that uses active resin containing solder fillers may allow reliable interconnects at low-cost. Fundamental process of Self-organization assembly are movement, coalescence, and wetting of molten fillers in resin. The focus of this paper is on the movement of fillers. In-situ observations of coalescence behavior for 40μm molten fillers in the resin revealed irregular movement of the fillers at a velocity of several μm/s. Numerical analysis, using improved volume fraction method, indicated that a 10% degradation of interfacial energy on one side of a 40μm filler could move the filler at a velocity of several mm/s. This degradation of the interfacial energy was resulted the remaining oxide film.

Interface Properties of Thin Film Bonding by Low Melting Point Metal for MEMS devices

Low temperature metal bonding is promising for assembling micro devices because the interface can easily achieve high mechanical reliability by the formation of metallurgical and electronically high conductive interfaces. For gaining high strength of micro joints with low temperature process, vacuum deposited low melting point metal (Snxln1-x) thin films were applied to solid-liquid diffusion bonding of copper. The effects of Sn content on the growth of reaction layer and on the shear strength were investigated. It was found that the sufficient initial Sn supply was important to achieve high strength interface characteristics to enhance solid-liquid diffusion.