Intelligent Parts Recognition and Navigation using Robotic Arms (IPRN)**

点击翻译按钮获取中文摘要

Parts production is a necessary process in manufacturing. The raw machine parts have to be identified and fitted into specific locations for cutting, drilling, grinding and so on. Currently, this process is mainly manual and labour intensive. Due to the high labour cost in North America, the parts production process is often out-sourced to cheaper labour markets outside Canada. Rational Robotics is an Edmonton based company, which specializes in developing robotic arms. The goal of this proposal is to automate the parts production process, using robotic operation and applying state-of-the-art human-computer interaction, computer vision, image analysis and pattern recognition techniques. Instead of identifying and inspecting parts manually, cameras will be deployed to capture multiple views. The captured images will be analyzed to match specific profiles in a parts database. The challenge is to recognize a unique part using minimum number of views in real-time. We will compare traditional vision-based algorithms with machine learning methods, and finally design an efficient approach to address this research problem. The proposed tasks will support a robotic automation pipeline: picking up an item from a pile of parts, viewing and recognizing, and fitting into the correct machine position. When the machining is done, an inspection step will be performed. A recovery procedure is available if needed. By embedding the ENGAGE outcome into Rational Robotics' already developed robotic arms, the company can accelerate a new market not only in parts production, but also other robotic applications. Without the need to out-source labour intensive jobs outside the country, Canada can benefit economically and technologically.