The purpose of this work is to explore the design principles for a Real-Time Robotic Multi\nCamera Vision System, in a case study involving a real world competition of autonomous driving.\nDesign practices from vision and real-time research areas are applied into a Real-Time Robotic Vision\napplication, thus exemplifying good algorithm design practices, the advantages of employing the\nââ?¬Å?zero copy one passââ?¬Â methodology and associated trade-offs leading to the selection of a controller\nplatform. The vision tasks under study are: (i) recognition of a ââ?¬Å?flatââ?¬Â signal; and (ii) track following,\nrequiring 3D reconstruction. This research firstly improves the used algorithms for the mentioned\ntasks and finally selects the controller hardware. Optimization for the shown algorithms yielded\nfrom 1.5 times to 190 times improvements, always with acceptable quality for the target application,\nwith algorithm optimization being more important on lower computing power platforms. Results\nalso include a 3-cm and five-degree accuracy for lane tracking and 100% accuracy for signalling panel\nrecognition, which are better than most results found in the literature for this application. Clear\nresults comparing different PC platforms for the mentioned Robotic Vision tasks are also shown,\ndemonstrating trade-offs between accuracy and computing power, leading to the proper choice of\ncontrol platform. The presented design principles are portable to other applications, where Real-Time\nconstraints exist.
Loading....