面向视觉惯性里程计的传感器数据同步采集方法
首发时间:2020-11-04
摘要:数据采集是视觉惯性里程计(Visual Inertia Odometry,VIO)算法实现的基础和前提,而面向VIO的数据采集通常需要解决视觉惯性数据的同步问题。传统的传感器数据软件采集方法无法实时地实现视觉与惯性数据的同步稳定采集。为解决该问题,在多进程编程基础上,提出面向视觉惯性里程计的传感器数据同步采集方法。该方法利用应用程序接口(Application Programmable Interface,API)编写数据采集程序,传感器通过通用串行总线(Universal Serial Bus, USB)与数据采集平台连接,可实现对视觉与惯性数据的同步稳定采集,无需进行离线数据同步。针对视觉数据采集对惯性数据采集产生干扰的问题,将两类数据的采集任务分为两个独立进程,并将每组数据的采集时间进行约束。试验结果表明,该数据采集方法可以实现对视觉和惯性数据的同步稳定采集。
关键词: 计算机科学与技术 视觉惯性里程计 数据采集 多进程编程 应用程序接口
For information in English, please click here
Sensor data synchronization acquisition method for visual inertial odometry
Abstract:Data collection is the basis and premise for the realization of the Visual Inertia Odometry (VIO) algorithm, and data collection for VIO usually needs to solve the problem of synchronization of visual inertia data. The traditional sensor data software collection method cannot realize the synchronous and stable collection of vision and inertial data in real time due to the large amount of visual data. To solve this problem, on the basis of multi-process programming, a method of sensor data synchronization for visual inertial odometer is proposed. This method uses the application program interface to write the data acquisition program, the sensor is connected to the data acquisition platform through the Universal Serial Bus (USB), and realizes the synchronous and stable acquisition of visual and inertial data without offline data synchronization. Aiming at the problem of visual data collection interfering with inertial data collection, the collection tasks of the two types of data are divided into two independent processes, and the collection time of each group of data is constrained. Finally, the experimental results show that the data collection method can realize the synchronous and stable collection of visual and inertial data.
Keywords: computer science and technology;visual Inertia odometry; data collection, multi-process programming application programming interface
引用
No.****
动态公开评议
共计0人参与
勘误表
面向视觉惯性里程计的传感器数据同步采集方法
评论
全部评论0/1000