【摘要】 为了提高采摘机器人成功避障的性能,在采摘机器人机器视觉技术中引入了DSP实时编码技术,通过对成熟果蔬图像和视频的实时编码,获取果蔬位置和障碍物位置,达到成功避障的目的。根据机器人视频采集驱动的软件架构,对采集驱动进行了改造,在不增加前端外设的前提下实现了主动降帧,并根据DSP芯片的特点,对实时编码器进行了优化设计,大大提高了高分辨率图像实时编码的速率。最后,对机器人的距离位置测量和避障功能进行了测试,通过测试发现:在1.20~2.00m的距离范围内,测量的距离和实际距离的最大偏差不超过0.03m,6组避障测试中机器人成功避障的概率较高,其中最低也达到了97.28%,符合设计的要求。更多还原

【Abstract】 In order to improve the picking robot successfully obstacle avoidance performance,it takes machine vision technology and DSP real- time encoding technology into picking robot. Through real- time encoding of mature fruit images and video,it can obtain the fruits and vegetables position and the obstacle position to successfully avoid obstacles. According to the software architecture for robot video capture driver,under the condition of not increasing front- end peripherals to achieve the initiative to drop frame,it drives the collection transformation. According to the characteristics of DSP chip,the real- time encoder of optimization was designed. It greatly increased rate of high resolution real- time image coding. Finally,the robot location of the distance measurement and obstacle avoidance capabilities were tested,it is found that within the distance is with the range of 1. 20- 2. 00 m. Measurement of the distance and the actual distance of the maximum deviation less than 0. 03 m,six groups of obstacle avoidance test robot successfully avoided obstacles with the higher probability,reached a minimum of 97. 28% and met the design requirements.更多还原