Stephen L. Chiu

Task compatibility of manipulator postures

In performing a manipulation task, humans tend to adopt arm postures that most effectively utilize the motion and strength capabilities of the arm. Selecting arm postures that are compatible with the task requirements has become almost instinctive to humans. By mimicking this approach in robotic manipulation, we can exploit the full capability of a manipu lator in performing a task. An index is proposed for measur ing the compatibility of manipulator postures with respect to a generalized task description. The manipulator is viewed as a mechanical transformer, with joint space velocity and force as input and task space velocity and force as output. Optimi zation of the index corresponds to matching the velocity and force transmission characteristics to the task requirements. The applications of this index to manipulator redundancy utilization and workspace design are also discussed.

Control of redundant manipulators for task compatibility

When called upon to perform a manipulation task, through experience we place our arm in a posture which is best suited for performing that task. Posture variation is a means through which the motion and strength characteristic of the arm is made compatible with the task requirements. We propose an index for measuring the compatibility of an arm posture with respect to fine manipulation tasks, where accurate control of small velocity and force is required, and coarse manipulation tasks, where exertion of large velocity and force is required. The optimization of this index presents an effective way of utilizing manipulator redundancy. The resultant manipulator postures closely resemble those used by humans in performing similar tasks.

A fuzzy logic programming environment for real-time control

Abstract Expert systems based on fuzzy logic inferencing have been shown to be effective in controlling complex processes. The experiences of human operators are naturally captured as linguistic fuzzy control rules. This paper describes a programming environment being developed to facilitate the implementation of fuzzy control systems that allow a user to easily describe a set of fuzzy rules, graphically edit the fuzzy variable definitions, and verify the rules through simulation. In an actual control system, the rules will reside in and be processed by a special-purpose chip for fuzzy logic inferencing. An object-oriented approach to the programming environment naturally models this hardware architecture. Each set of rules is associated with a chip object simulated in software. The content of the chip object is changed as the rules are modified. When the simulation results are satisfactory, the content of the simulated chip can be copied directly into the hardware chip for real-time applications. A complex controller that involves multiple domains of expertise can be developed by first focusing on component chip objects and then interconnecting the components. The ability to translate linguistic rules into practical implementations is a unique and useful feature of this environment.

A DEVELOPMENT ENVIRONMENT FOR FUZZY RULE-BASED TRAFFIC CONTROL

We present a development environment for fuzzy rule-based traffic control systems. This Macintosh® environment allows a user to simulate traffic on a network of streets and create fuzzy decision rules for adjusting traffic signal timing parameters at the intersections. During simulation, various graphical displays help the user interactively debug and fine-tune the decision rules. When the results are satisfactory, C or Ada code can be automatically generated to implement the fuzzy knowledge base. We also describe a specific knowledge base developed using this environment and present some results.