AWARE

VERTLINER is developing aerial robotic platforms to improve waste management in indoor construction sites. Our UAVs detect and categorise waste using advanced sensors, producing rich data that could be subject to train neural-based powerful AI models with EOSC DIH’s computing resources. After thorough field testing and feedback, we continuously refine our approach, emphasising both technological progress and real-world construction solutions.

VERTLINER is a pioneering field robotics company specialising in the creation of autonomous robotic aerial platforms for indoor asset assessments. Our newest pilot, AWARE, leverages a semi-autonomous UAV equipped with advanced sensor modalities. This software’s purpose is to detect materials frequently seen on construction sites. By identifying these materials, we aim to significantly improve waste management processes.

Challenge

Effective waste management in construction sites necessitates precise identification and location of various materials. Traditional methods can be time-consuming and often inaccurate. Through the use of High-Performance Computing (HPC) services, the AWARE pilot seeks to harness the power of AI by training a Neural Network based on the DIH cloud resources such as GPUs, CPUs, RAM, and disk space. This will enable us to develop an efficient AI model for accurate material detection, ultimately enhancing VERTLINER’s information gathering capabilities regarding objects in areas of interest.

Work Plan

Our pilot kicks off with data acquisition using our semi-autonomous UAV, which will survey a variety of construction sites. Once collected, this data undergoes rigorous processing and labelling, preparing it for the training phase. We’ll then harness High-Performance Computing (HPC) & Cloud services to train and refine our AI model, ensuring its accuracy and efficacy on real-world data.

Following the development phase, this AI model will be seamlessly integrated into VERTLINER’s advanced aerial platforms. Field tests will be conducted in actual construction settings, validating the model’s performance and reliability. Post-testing, we’re committed to making iterative enhancements, drawing from invaluable data insights and on-the-ground feedback, ensuring the continual advancement of our solution.