| Summary: | 博士 === 國防大學理工學院 === 國防科學研究所 === 105 === Mini-Unmanned Aerial Vehicle (Mini-UAV) and Micro Aerial Vehicle (MAV) have already been widespread utilization in military and livelihood, because the flying self-control with navigation technology is much improved nowadays. Nevertheless, the flying altitude in the application of Mini-UAV to carry out a task is not highly demanded; therefore, the accidental collisions are inevitable as any obstacle on the ground. Additionally, the Mini-UAV, which is confined to equip radar or laser distance gauge, is unable to have obstacle avoidance function. Thus, this research proposes an obstacle avoidance system, which conforms to the Mini-UAV and originates in visual images as the detection, may detect the obstacle in front of the flight path and control the flight vehicle to take a reaction of dodge. This study proposes obstacle avoidance algorithm based on vision detection. The stable flight is achieved by skyline detection in an image captured by a camera mounted on a Mini-UAV, which is combined with obstacle detection to avoid obstacles in the flight path. The avoidance algorithm is calculated by using the information related to the skyline and obstacles using a ground computer, which facilitates stable, horizontal flight with obstacle avoidance. The obstacle avoidance algorithm is completed by using the flight control system and a program interface is developed, which combine the hardware interface of a digital-to-analog converter, a video capture card, the Mini-UAV remote controller, and other related devices. Experimental results demonstrate that the practical utility of the vision-based flight control method for obstacle avoidance in a virtual environment can be reached effectively.
|
|---|
