Safe robot navigation and infrastructure data capture within commercial environments via optimal positioning of sensors

点击翻译按钮获取中文摘要

Capital projects employ engineers to design assets based on project plans. Parts are built in factories all overthe world and shipped to site which in numerous occasions do not fit to the existing infrastructure. These andother factors cost the capital project industry $1.6 Trillion annually. Using a revolutionary and comprehensiveend-to-end software solution to solve the problem of industrial rework and facilitate proactive, cost effectiveproject execution digital project information generated via the collection of a variety of sensory and spatial datato create a Digital Twin of the physical project is currently being used. However, data collection, typicallyperformed manually, requires to be automated to enable effective capital project execution while maintainingthe required sensor data accuracy.By identifying mismatches between reality and plan, the automated sensor data collection will enable correctiveactions early in the project execution before they cause rework.The focus of this project is to automate capital project fast data collection in high definition via the use ofautonomous Unmanned Ground Vehicles (UGVs).In particular we will develop enhanced robot's obstacle avoidance mechanisms to better estimate sensorplacement and maneuvering the UGV to optimally position the sensors for relevant and accurate datacollection. A vision input (LIDAR sensor) will be used as a feedback to servo the UGV for 3D motionpurposes while having the capabilities to quickly identify precise sensor placement. The focus is on developingtools for (1) accurate detection of complex obstacles typically found in commercial sites (e.g., overhandcables), (2) UGV path planning for optimal sensor placement via visual servoing/surveying, and (3) analyze theeffectiveness of the use of UGVs in construction sites.