Fan-Tien Cheng

Efficient algorithm for optimal force distribution-the compact-dual LP method

An efficient algorithm, the compact-dual linear programming (LP) method, is presented to solve the force distribution problem. In this method, the general solution of the linear equality constraints is obtained by transforming the underspecified matrix into row-reduced echelon form; then, the linear equality constraints of the force distribution problem are eliminated. In addition, the duality theory of linear programming is applied. The resulting method is applicable to a wide range of systems, constraints, and objective functions and yet is computationally efficient. The significance of this method is demonstrated by solving the force distribution problem of a grasping system under development at Ohio State called DIGITS. With two fingers grasping an object and hard point contact with friction considered, the CPU time on a VAX-11/785 computer is only 1.47 ms. If four fingers are considered and a linear programming package in the IMSL library is utilized, the CPU time is then less than 45 ms. >

Resolving manipulator redundancy under inequality constraints

Due to hardware limitations, physical constraints such as joint rate bounds, joint angle limits, and joint torque constraints always exist. In this paper, these constraints are considered in the general formulation of the redundant inverse kinematic problem. To take these physical constraints into account, the computationally efficient compact quadratic programming (QP) method is formed to resolve the constrained kinematic redundancy problem. In addition, the compact-inverse QP method is also formulated to remedy the inescapable singularity problem with inequality constraints. Two examples are given to demonstrate the generality and superiority of these two methods: to eliminate the drift phenomenon caused by self motion and to remedy saturation-type nonlinearity problems. >

Optimal force distribution in multiple-chain robotic systems

The force-distribution problem in multiple-chain robotic systems is to solve for the setpoints of the chain contact forces and input joint torques for a particular system task. It is usually underspecified, and an optimal solution may be obtained. The generality of the compact-dual linear programming (LP) method that can accept a variety of linear objective functions for different applications over a wide range of multiple-chain systems (multilegged vehicles, dexterous hands, and multiple manipulators) is demonstrated; and the solutions for several common problems of force distribution including slippage avoidance, minimum effort, load balance, and temporal continuity are proposed. This is illustrated by solving the force-distribution problem of a grasping system being developed called Digits. Efficiency considerations and elimination of redundant constraints are also discussed. With four fingers grasping an object, considering a conservative friction coefficient (for safety margins on friction constraints) and using a combined objective function for achieving the goals of minimum effort, load balance, and temporal continuity, the CPU time on a VAX-11/785 computer is less than 45 ms (using a linear programming package in the IMSL library). Therefore, it is believed that rather general use of the compact-dual LP method may be made to define a suitable objective function for a particular application and to solve the corresponding force-distribution problem in real time. >

Study and resolution of singularities for a 6-DOF PUMA manipulator

Upon solving the inverse kinematics problem of robot manipulators, the inherent singularity problem should always be considered. When a manipulator is approaching a singular configuration, a certain degree of freedom will be lost such that there are no feasible solutions of the manipulator to move into this singular direction. In this paper, the singularities of a 6-DOF PUMA manipulator are analyzed in detail and all the corresponding singular directions in task space are clearly identified. In order to resolve this singularity problem, an approach denoted Singularity Isolation Plus Compact QP (SICQP) method is proposed. The SICQP method decomposes the work space into achievable and unachievable (i.e., singular) directions. Then, the exactness in the singular directions are released such that extra redundancy is provided to the achievable directions. Finally, the Compact QP method is applied to maintain the exactness in the achievable directions, and to minimize the tracking errors in the singular directions under the condition that feasible joint solutions must be obtained. In the end, some simulation results for PUMA manipulator are given to demonstrate the effectiveness of the SICQP method.

A Novel Virtual Metrology Scheme for Predicting CVD Thickness in Semiconductor Manufacturing

In an advanced semiconductor fab, online quality monitoring of wafers is required for maintaining high stability and yield of production equipment. The current practice of only measuring monitor wafers may not be able to timely detect the equipment-performance drift happening in-between the scheduled measurements. This may cause defects of production wafers and, thereby, raise the production cost. In this paper, a novel virtual metrology scheme (VMS) is proposed for overcoming this problem. The proposed VMS is capable of predicting the quality of each production wafer using parameters data from production equipment. Consequently, equipment-performance drift can be detected promptly. A radial basis function neural network is adopted to construct the virtual metrology model. Also, a model parameter coordinator is developed to effectively increase the prediction accuracy of the VMS. The chemical vapor deposition (CVD) process in semiconductor manufacturing is used to test and verify the effectiveness of the proposed VMS. Test results show that the proposed VMS demonstrates several advantages over the one based on back-propagation neural network and can achieve high prediction accuracy with mean absolute percentage error being 0.34% and maximum error being 1.15%. The proposed VMS is simple yet effective, and can be practically applied to construct the prediction models of semiconductor CVD processes.

Development of a System Framework for the Computer-Integrated Manufacturing Execution System：a Distributed Object-Oriented Approach

Today, most of the Manufacturing Execution Systems (MES) are monolithic, insufficiently configurable and difficult to modify. Using a distributed object-oriented technique, we will present a systematic approach, to develop a computer-integrated MES Framework which is open, modularized, distributed, configurable, interoperable, and maintainable. We start with the system analysis by collecting domain requirements and analysing domain knowledge. Our MES Framework is designed by the process of constructing an abstract object model based on domain knowledge, partitioning the application domain into components, identifying generic parts among components, defining framework messages, and developing design patterns for generic parts. After finishing the MES Framework design, we design a variety of functional components by inheriting appropriate design patterns of the Framework. An application can then be constructed by invoking corresponding methods of related components. The proposed MES Framework can be integra...

Dual-Phase Virtual Metrology Scheme

This paper proposes a dual-phase virtual metrology scheme. To consider both promptness and accuracy, this scheme generates dual-phase virtual metrology (VM) values. Phase I emphasizes promptness, that is to immediately calculate and output the Phase-I VM value (denoted ) of a workpiece (wafer or glass) once the entire process data of the workpiece are completely collected. Phase II improves accuracy, that is not to recalculate and output the Phase-II VM values (denoted ) of all the workpieces in the cassette (also called FOUP in the semiconductor industry) until an actual metrology value (required for tuning or retraining purposes) of a workpiece in the same cassette is collected. Also, in this scheme, the accompanying reliance index (RI) and global similarity index (GSI) of each and are also generated. The RI and GSI are applied to gauge the degree of reliance. If the reliance level of a VM value is lower than the threshold, this VM value may not be adopted. An illustrative example involving fifth-generation thin-film transistor liquid crystal display (TFT-LCD) chemical-vapor deposition equipment is presented. Experimental results demonstrate that the proposed scheme is applicable to the wafer-to-wafer or glass-to-glass advanced process control for semiconductor or TFT-LCD factories.

Application development of virtual metrology in semiconductor industry

Daily wafer fabrication in semiconductor foundry depends on considerable metrology operations for tool-quality and process-quality assurance. The metrology operations required a lot of metrology tools, which increase FABs investment. Also, these metrology operations will increase cycle time of wafer process. Metrology operations do not bring any value added to wafer but only quality assurance. This article provides a new method denoted virtual metrology (VM) to utilize sensor data collected from 300 mm FABs tools to forecast quality data of wafers and tools. This proposed method designs key steps to establish a VM control model based on neural networks and to develop and deploy applications following SEMI EDA (equipment data acquisition) standards.

The improved compact QP method for resolving manipulator redundancy

The compact QP method is an effective and efficient algorithm for resolving the manipulator redundancy under inequality constraints. In this paper, a more computationally efficient scheme which will improve the efficiency of the compact QP method-the improved compact BP method-is developed. With the technique of work space decomposition, the redundant inverse kinematics problem can be decomposed into two subproblems. Thus, the size of the redundancy problem can be reduced. For an n degree-of-freedom spatial redundant manipulator, instead of a 6/spl times/n matrix, only a 3/spl times/(n-3) matrix is needed to be manipulated by Gaussian elimination with partial pivoting for selecting the free variables. The simulation results on the CESAR manipulator indicate that the speedup of the compact QP method as compared with the original QP method is about 3.3. Furthermore, the speedup of the improved compact QP method is about 5.6. Therefore, it is believed that the improved compact QP method is one of the most efficient and effective optimization algorithm for resolving the manipulator redundancy under inequality constraints. >

Efficient formulation of the force-distribution equations for simple closed-chain robotic mechanisms

Force distribution is the inverse dynamics problem for multiple-chain systems in which the motion is completely specified and the internal forces/torques to effect this motion are to be determined. A computationally efficient formulation for the force-distribution problem is presented. This formulation is applicable to a number of simple closed-chain robotic mechanisms, including dexterous hands, multiple manipulators, and multilegged vehicles. Modeling of chain contacts is relatively general so that hard point contact, soft finger contact or rigid contact with an irregularly-shaped object or with uneven terrain may be handled. The dynamic effects of the chains and physical limits on their actuators are efficiently included in the formulation through the use of the inverse dynamics and Jacobian relationships for each chain. Based on this efficient formulation, a variety of methods may then be developed to solve the force-distribution problem. >