Learning-based autonomous robotic system for package sorting application

随着电子商务和物流行业的出现，每天都有大量的货物被交付。COVID-19疫情导致的封锁迫使企业和客户使用电子商务系统在线销售或购买产品和服务。分拣系统在订单交付实践中发挥着关键作用。通常，分拣系统由捆绑在一起的输送机组成，这些输送机由仓库管理系统（WMS）控制。传统的物品分拣线依靠人工操作员手动定位和拾取每个产品。这种系统是劳动密集型的，耗时的，并且容易出错，导致定期进行多次计数。为了解决这些问题，已经采用了自动分拣系统。尽管以前的努力，大多数机器人分拣系统无法在复杂的环境中工作，现实生活中的应用。在这些情况下，机器人应该能够准确地识别具有各种物理特征的物体，并快速对其进行分类。为此，自主分拣机器人应具有以下能力：（i）检测和分类具有不同形状，尺寸和物理特性的物体，（ii）最佳物体抓取，以及（iii）3D环境中的轨迹生成和运动规划。为了满足这些要求，我们将联合收割机最先进的计算机视觉算法与人工智能方法相结合，开发一种自动机器人系统，用于在现实环境中分拣包裹。该项目将在以下方面使加拿大受益匪浅。(i)加拿大的许多行业将从拟议的自动分拣系统中受益。该项目的成果对生产、仓储、库存规划和运输部门产生了积极影响。(ii)该项目对最先进技术的需求将增加加拿大高素质人员（HQP）的就业机会，并为实际工程应用培训他们。(iii)该项目将促进加拿大电子商务和物流业的发展，以增加其在全球市场的市场份额。

With the advent of e-commerce and logistics industry, numerous goods are delivered every day. The COVID-19 pandemic-driven lockdown has forced businesses and customers to use e-commerce systems in order to sell or purchase products and services online. Sorting systems play a pivotal role in orders delivery practices. Typically, sorting systems consist of a mix of conveyors tied together which are controlled by the Warehouse Management System (WMS). Traditional objects sorting lines rely on human operators to locate and pick each product by hand. Such systems are labour-intensive, time-consuming, and error-prone, resulting in multiple counts to be done on a regular basis. To remedy these problems, automated sorting systems have been adopted. Despite the previous efforts, most robotic sorting systems fail to work in a complex environment for real-life applications. In these situations, robots should be able to recognize objects with various physical characteristics accurately and sort them quickly. To this end, an autonomous sorting robot should have the following capabilities: (i) detection and classification of objects with different shapes, sizes and physical properties, (ii) optimal object grasping, and (iii) trajectory generation and motion planning within the 3D environment. To fulfill these requirements, we will combine state-of-the-art computer vision algorithms with artificial intelligence methods to develop an autonomous robotic system for sorting packages in a real-world environment. This project will considerably benefit Canada in the following ways. (i) A wide range of Canada's industries will benefit from the proposed automated sorting system. The outcomes of this project have positive impacts on production, warehousing, inventory planning, and transportation sectors. (ii) The need for state-of-the-art technologies in this project will increase job opportunities for High Quality Personnel (HQP) in Canada and train them for real-world engineering applications. (iii) This project will precipitate the growth of Canada's e-commerce and logistic industry to increase its market share in the global market.