Abstract: Using an embedded EPXA10 as the core and an automated sewage discharge robot as the application, this paper introduces a binocular vision positioning system. A method is proposed that utilizes fuzzy algorithms for edge detection of target images, combined with binocular vision positioning algorithms to achieve machine vision positioning. Considering the working characteristics of the sewage discharge robot, an active method of finding special images is adopted to achieve rapid and accurate target tracking and positioning, completing the three-dimensional positioning of the sewage discharge port. Based on the positioning coordinates, the robot can accurately connect the sewage discharge pipe and quickly discharge sewage into the pool.
Keywords: Embedded systems, robots, vision positioning systems
Abstract: Based on the embedded processor EPXA10, a system of Binocular Stereo Vision Localization for a robot system of sewage lorry self-discharging is presented in this paper. Especially, a new method of robot localization is presented, in which the Fuzzy theory is adopted for the edge detecting of the target image and the Stereo Vision algorithm is used for localization. Aimed at the special role of the robot, a method of actively searching a special image-symbol which can achieve a fast and exact localization is adopted for the purpose of three-dimension localization of the nozzle on the sewage lorry. According to the located coordinate, the robot can connect the sewage discharging pipes, and let sewage into the puddle.
Key words: Embedded system; Binocular Stereo localization; Robot vision; Sewage Discharge; Robot
1 Introduction
Robot vision localization systems are a crucial part of robotics. This paper focuses on a sewage discharge robot, introducing a vision localization system for such a robot based on an embedded system. The paper employs a binocular vision localization algorithm to locate the robot and utilizes a passive vision-based camera autonomous calibration method. The camera's perspective projection matrix is obtained using a linear least squares algorithm for calibration. Finally, a special image marker guides the robot to actively seek out and track the target.
As a water-scarce country, my country's wastewater treatment systems generally suffer from insufficient automation. Before treatment, wastewater needs to be discharged from transport vehicles into wastewater tanks. Currently, this discharge is mostly done manually. The wastewater discharge robot mentioned in this article aims to replace humans in this task, accurately connecting the wastewater inlet pipe of the wastewater tank to the pipe on the wastewater truck. This has significant social implications and practical application prospects (as shown in Figure 1). The system composition is shown in Figure 1:
Figure 1 Schematic diagram of sewage discharge system
A binocular vision localization algorithm is used to locate the robot, and a passive vision-based camera autonomous calibration method is used. The camera perspective projection matrix is obtained by using a linear least squares algorithm to calibrate the camera. Finally, a special image marker is used to guide the robot to actively find the target and complete the target localization and tracking.
2. Introduction to Embedded System Image Processors
Image processing often involves multiple steps, such as preprocessing, detection, tracking, and recognition (as shown in Figure 1), resulting in a very large amount of data required for real-time image processing. Furthermore, embedded processors used for image analysis and processing have extremely high real-time performance requirements. This paper utilizes Altera's EPXA10 chip, which integrates high-density logic (FPGA), memory (SRAM), and an embedded processor (ARM) onto a single programmable logic device, achieving a perfect combination of speed and programmability.
For details, please click: Research on Robot Vision Positioning System Based on Embedded System
