Application Research Of Indoor Obstacles Avoidance About Quadcopter Basded On QCOS/JVM

Quadcopter is a compact and flexible Mini-UAV. It have the characteristics of a vertical take-off and landing, hovering, flying depend on small space,equipped with a variety of sensors, quadcopter widely applied research development. So,Quadcopter application research has been extensively developed.The main innovation of this thesis is embodied in that after analyzing quadcopter flight control task composition, it implements quadcopter operating system QCOS in order to improve the efficiency of task scheduling. And it established JAVA virtual machine and high-performance QCOS/JVM software platform based on the operating system. This thesis achieved quadcopter obstacle avoidance indoor with single camera on high-performance software platform.Firstly, building quadcopter visual system platform to solve the quadcopter interior avoidance complex problems and to reduce interference.Secondly, on hardware platform of the quadcopter.we designed and implemented quadcopter operating system QCOS, which is focusing on quadcopter flight control task scheduling features and following Northeastern University's independent research and development of embedded technology laboratory IDCX? embedded operating system as prototype. We proposed an improved task scheduling algorithm,which is suitable to quadcopter task scheduling, to reduce the overhead of task preemption and to enable it a high real-time, low-latency characteristics. And it has been verified in terms of real-time execution capabilities and effectiveness.In addition, this thesis implemented a streamlined version of the JAVA virtual machine based on real-time systems QCOS that quadcopter is running, and mainly designed real-time task execution mechanism of JAVA applications and the corresponding class constructor. The high performanc QCOS/JVM software platform, which is formed by a joint optimization of quadcopter operating system QCOS designed and JAVA virtual machine implemented, can be fully used for real-time services JAVA image acquisition and control signal processing.Finally, this thesis implemented obstacle avoidance in the open space interior based on the high-performance QCOS/JVM. Taking advantage of the principle of pyramid LK optical flow, I gained the actual optical flow and the translation of optical flow due to the motion of four axes integrated optical flow algorithm was proposed in order to identify moving obstacle. Actual flight shows the test and comparison on the real-time ability and accuracy of merged optical flow algorithm this thesis proposed.