The allocation of hard real-time tasks to heterogeneous processors in robotic and digital control systems

The authors present a systematic approach to modeling the timing behavior of hard real-time (HRT) control algorithms. Parallel computers have increasingly been used in the implementation of large digital control systems. The dynamic environment surrounding complex HRT systems requires that the parallel computer architecture be flexible and able to support different technologies, including heterogeneous processor sets. A general formulation of scheduling models for digital and robotic control systems is described. A scheduling tool is presented for managing the complexity of nonpreemptive scheduling and allocation of HRT tasks to homogeneous and heterogeneous processor sets. The scheduling algorithms and the implementation of the scheduling tool are discussed. The test cases used to validate the proposed algorithms and the results obtained from these tests are presented.<
>