Title :
Vision based ground marker fast detection for small robotic UAV
Author :
Anping Gu ; Jinfa Xu
Author_Institution :
Dept. of Aircraft, Nanjing Univ. of Astronaut. & Aeronaut., Nanjing, China
Abstract :
A vision based real-time circular ground marker detection method, which is used to guild small autonomous control robotic UAV(RUAV) to pick up ground target object, is presented. The method is based on RHT algorithm and running entirely on GPU via Microsoft DirectX 11 API, since the CPU onboard doesn´t have sufficient computing power. In order to get a highest possible performance, special efforts are made in many ways, such as threads workloads balancing, data transfer lag hiding, etc. Experimental result shows the proposed method can give proper detection result and reach a high enough frame rate on a notebook level GPU, running much faster than the CPU version of detection method.
Keywords :
autonomous aerial vehicles; graphics processing units; microrobots; object detection; robot vision; CPU onboard; GPU; Microsoft DirectX 11 API; RHT algorithm; RUAV; data transfer lag hiding; ground target object detection; notebook level GPU; small autonomous control robotic UAV; small robotic UAV; sufficient computing power; thread workload balancing; vision based ground marker fast detection; vision based real-time circular ground marker detection method; Computers; Data transfer; Graphics processing units; Image edge detection; Instruction sets; Robots; Transforms; DirectX 11; GPU; circle detection; robotic vision;
Conference_Titel :
Software Engineering and Service Science (ICSESS), 2014 5th IEEE International Conference on
Conference_Location :
Beijing
Print_ISBN :
978-1-4799-3278-8
DOI :
10.1109/ICSESS.2014.6933728