Deep learning of grasping detection for a robot used in sorting construction and demolition waste
- PDF / 1,940,740 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 53 Downloads / 225 Views
ORIGINAL ARTICLE
Deep learning of grasping detection for a robot used in sorting construction and demolition waste Yuedong Ku1 · Jianhong Yang1 · Huaiying Fang1 · Wen Xiao1 · Jiangteng Zhuang1 Received: 18 November 2019 / Accepted: 23 August 2020 © Springer Japan KK, part of Springer Nature 2020
Abstract The traditional construction and demolition waste (CDW) treatment process adopts the method of crushing and screening after mixing and combines the method with manual sorting for resource recycling. However, there is a problem of low recycling purity and low efficiency of manual sorting after mixed screening. This paper proposes a robot for sorting CDW, which is used to finely sort a large number of objects before mixing and crushing. The use of the robot improves the level of resource utilization of CDW. However, under actual working conditions, the adhesion and stacking of CDW on the conveyor belt and the irregularity of the shapes of CDW lead to errors in grasping-information. Thus, a deep learning method for grasping detection is proposed. The method generates some grasping rectangles through a searching algorithm, and inputs the rectangles to the neural network. Then, the network outputs the optimal grasping pose. The experiment demonstrated that the original accuracy of robotic grasping was only 70%. After deep learning for grasping detection, the accuracy was over 90%, which thoroughly meets the requirements of efficiency and accuracy for sorting CDW under actual working conditions. Keywords Construction and demolition waste · Robotic sorting · Deep learning · Grasping detection
Introduction Necessity of robotic sorting In the modernization and urbanization of daily life, a large amount of construction and demolition waste (CDW) is inevitably generated. The European construction industry produces 820 million tons of CDW annually, which accounts for 46% of the total waste generated according to Eurostat [1, 2]. CDW essentially has the characteristics of a large number of objects, wide variety, complex composition, and low recycling rate. Traditional open-air stacking and landfill have seriously affected the environment [3, 4]. In China, the amount of CDW is the highest in the world, but the utilization rate is less than 5%. To achieve sustainability, the Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10163-020-01098-z) contains supplementary material, which is available to authorized users. * Jianhong Yang [email protected] 1
College of Mechanical Engineering and Automation, Huaqiao University, No. 668, Jimei Street, Xiamen City 361021, Fujian Province, China
government has begun to encourage enterprises to recycle CDW. Recycling methods can be divided into direct and indirect sorting by sensors. The direct sorting method includes crushing and screening, magnetic separation [5], light–matter separation [6], and triboelectrostatic separation [7]. The traditional method of recycling CDW is shown in Fig. 1. The traditional method mainly includes mixe
Data Loading...
