Distributed architecture and fast parallel algorithms in real-time robot control
Control algorithms for accurate and fast moving robots require the use of the complete dynamic model of the robot. The complete dynamic model takes into account the nonlinearity and coupling terms of the whole structure. The time of computing the dynamic equations determines whether they are usable in practical applications. Reducing the computation time can be achieved by simplifying the dynamic model, simplifying the calculations, and improving the computer's computation speed. The proposed solution is a distributed computer architecture, and new fast parallel algorithms which allow efficient utilization of the computers. This architecture has several advantages: modularity, reliability, spatial distribution, and speed. Because of the serial nature of the dynamic calculations, it is difficult to achieve efficient utilizations of the multiple computers. With the new algorithms, motion, and force values, which must be passed between the computers, are replaced by predicted values. The predicted information is acquired from values calculated in previous sample intervals. This allows efficient use of the computers and increases the computation speed of the control algorithms.
- Research Organization:
- Oregon State Univ., Corvallis (USA)
- OSTI ID:
- 5790836
- Country of Publication:
- United States
- Language:
- English
Similar Records
Hierarchical multiprocessor architecture design in VLSI for real-time robotic control applications
Modeling and simulation of robot dynamics using transputer-based architectures