Pipelined multiprocessor computer architecture and fast parallel algorithms for real time robot control
As a result of the automation revolution, robots are assuming ever more complex and demanding tasks. Robot system control schemes for fast and precise robot motion require utilization of the entirety of the robot dynamic formulations and the ability to evaluate these formulations in real-time. The dynamic formulations, which take into account robot nonlinearities and dynamic coupling, are computationally intensive. They are difficult to implement in real-time at high sampling rates due to the time required to compute the dynamic formulations. Reducing the computation time for practical implementation can be achieved by developing a computing algorithm for the efficient evaluation of the dynamic formulations and by designing a dedicated computer architecture. The proposed solution is a pipelined multiprocessor computer architecture and fast parallel algorithms for real-time control. The multiprocessor system can be utilized to concurrently perform pipelined parallel computations, thereby substantially increasing controller processing speed and CPU utilization. Concurrent performance of pipelined parallel computations is based on consideration of the sequential dependencies of the dynamic formulations which are conductive to pipelining, and decomposition of the dynamic backward formulations for fast parallel computation. The performance of the proposed algorithms, called PAFP, is evaluated through analytic error analysis and experimental simulations, including motion simulations. It is compared to other approaches to the problem proposed by, respectively, Bejczy and Binder.
- Research Organization:
- Oregon State Univ., Corvallis, OR (USA)
- OSTI ID:
- 5814942
- Resource Relation:
- Other Information: Thesis (Ph. D.)
- Country of Publication:
- United States
- Language:
- English
Similar Records
Parallel algorithms and architectures for the manipulator inertia matrix
Data Locality Enhancement of Dynamic Simulations for Exascale Computing (Final Report)
Related Subjects
ARRAY PROCESSORS
COMPUTER ARCHITECTURE
ROBOTS
CONTROL SYSTEMS
ALGORITHMS
COMPARATIVE EVALUATIONS
COST
DESIGN
ERRORS
NONLINEAR PROBLEMS
PARALLEL PROCESSING
PERFORMANCE
REAL TIME SYSTEMS
VECTOR PROCESSING
MATHEMATICAL LOGIC
PROGRAMMING
990200* - Mathematics & Computers