A learning force control strategy for robotic gripping and a mobile robot inspector system for intelligent factories

点击翻译按钮获取中文摘要

Four graduate and six undergraduate students in the Department of Mechanical Engineering at the University of New Brunswick will conduct research on this Discovery Grant project. One aspect of the research will focus on making robots more dexterous and skilled at handling varied objects. This can result in them perform tasks at higher levels of efficiency. Gripping is one of human's most-performed actions; however it is currently a challenge for robots to do this in a human-like way that is very efficient and dexterous. Gripping performed by humans is based on numerous known characteristics of objects and instantaneous decision making about the objects. The proposed research will target this challenge of making robots grip in a human-like way, working with 2 MScEs, 1 PhD and 4 undergraduates to derive an intelligent gripping protocol using flexible fingers on a robot. As many components can form a product, any one or more that is not the correct quality standard will result in an overall failed product. Plastic products constitute a vast number of variations in every day life. This is so since their manufacture can be conducted at high rates at a very low cost. While this is good to the end-user for diversity, making an acceptable product (automobile, medical, aerospace, appliances, packaging etc.) requires skilled personnel to identify the many defect metrics (visually) for product quality classifications when it come to plastic products. In terms of part dimensions, various measuring methods are used to scan the 3-D shape of parts for verifying dimensional quality. However, dimensional quality is time-varying. To compound this, inspection is conducted sparsely, once at a specific time, which may not capture the defect, resulting in many more failed products. The proposed research will also target this challenge, working with 1 PhD and 2 undergraduates to create a mobile inspector robot system (MRIS) that can perform inspection tasks in an on-demand manner with intelligence to advise other machines to make adjustments in the loop. There is significant benefit to Canadian industries in making robots more efficient at handling varied goods. The societal impact will also realize high value as robots become more intelligent at gripping. Having an intelligent MIRS will help identify low quality parts early so as to mitigate thousands of future rejects. Therefore, the impact on energy requirements for recycling can be significantly reduced, as well as reducing negative effects on the environment.