DEUCON: Decentralized End-to-End Utilization Control for Distributed Real-Time Systems
IEEE Transactions on Parallel and Distributed Systems
| Hi-index | 0.00 |
In vision-based feedback control systems, the time to obtainsensor information is usually non-negligible, and thesesystems thereby possess fundamentally different timing behaviorcompared to standard real-time control applications.For many image-based tracking algorithms, however, it ispossible to trade-off the computational time versus the accuracyof the produced position/orientation estimates.This paper presents a method for optimizing the use ofcomputational resources in a multi-camera based positioningsystem. A simplified equation for the covariance of theposition estimation error is calculated, which depends onthe set of cameras used and the number of edge detectionpoints in each image. An efficient algorithm for selectionof a suitable subset of the available cameras is presented,which attempts to minimize the estimation covariance givena desired, pre-specified maximum input-output latency ofthe feedback control loop.Simulations have been performed that capture the real-timeproperties of the vision-based tracking algorithm andthe effects of the timing on the performance of the controlsystem. The suggested strategy has been compared withheuristic algorithms, and it obtains large improvements inestimation accuracy and performance for objects both infree motion and under closed-loop position control.