<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><meta http-equiv=Content-Type content="text/html; charset=iso-8859-1"><meta name=Generator content="Microsoft Word 15 (filtered medium)"><style></style></head><body lang=PT link="#0563C1" vlink="#954F72"><div class=WordSection1>Call for Tenders for PhD Research Grant – Ref. BD4/2023/Topic 2 - Integration of 6D object pose estimation and environment representation with motion planning strategies for fine object grasping using mobile robots.<o:p></o:p>The Institute of Systems and Robotics (ISR-UC) is calling for applications for ONE PhD Research Grant in the area of Electrical Engineering and Intelligent Systems, or Electrical and Computer Engineering, under the FCT Research Grant Regulations (RBI) and the Research Grant Holder Statute (EBI).<o:p></o:p>The scholarships will be funded by the Foundation for Science and Technology (FCT) under the Collaboration Protocol for Funding the Multi-Year Plan for Research Scholarships for PhD Students, signed between the FCT and the R&D Unit, Institute of Systems and Robotics, UID/EEA/00048/2020.<o:p></o:p><o:p> </o:p>The aim is to explore object detection in an industrial environment and subsequent pick-and-place operations performed by a mobile robot. Machine learning methods will be applied to process multimodal data in robotic perception components and reinforcement learning to learn actions. The work plan consists of the following tasks:<o:p></o:p><o:p> </o:p><![if !supportLists]>1.       <![endif]>Research into 6D object pose estimation based on visual and/or depth sensor data. Deep Learning-based approaches have shown superior performance in object segmentation and point matching and general robustness, while geometric approaches tend to be more accurate (more refined optimisation) and more efficient. In this task, a hybrid approach will be followed.<o:p></o:p><o:p> </o:p><![if !supportLists]>2.       <![endif]>Specialisation of models in warehouse/industrial contexts. In particular, the detection of pallets and the estimation of their 6D pose, in an industrial context, for navigation applications (pick-and-place) using robotic platforms (e.g. forklifts).<o:p></o:p><o:p> </o:p><![if !supportLists]>3.       <![endif]>Research into navigation using reinforcement learning on the best behaviour a robotic platform should have when approaching objects for loading, including possible risk situations/scenarios and behaviours to minimise location errors. The aim is to use 6D object information to improve the robot's state representation (afordance-based and representation/image-based), so that it learns behaviours that avoid collisions, guarantee the platform's safety and allow it to autonomously choose the best docking/picking behaviours.<o:p></o:p><o:p> </o:p><![if !supportLists]>4.       <![endif]>Exploration and experimental evaluation in an indoor context and in a semi-structured context with the aim of validating the proposed methodologies.<o:p></o:p><o:p> </o:p>Supervisors: Prof Urbano Nunes, Prof João Barreto, Prof Rui Araújo.<o:p></o:p><o:p> </o:p>The competition is open between 1 October 2023 and 23:59 (Lisbon time) on 15 October 2023.<o:p></o:p>Applications and supporting documents must be submitted by email to <a href="mailto:lara@isr.uc.pt">lara@isr.uc.pt</a>, quoting reference BD4/2023/Topic 2. <o:p></o:p><o:p> </o:p>For further information please see: <a href="https://www.euraxess.pt/jobs/148028">https://www.euraxess.pt/jobs/148028</a> <o:p></o:p><o:p> </o:p>Cumprimentos,<o:p></o:p><o:p> </o:p>Lara Costa<o:p></o:p>Gestora de Ciência<o:p></o:p>Instituto de Sistemas e Robótica<o:p></o:p>Departamento de Engenharia Eletrotécnica e de Computadores<o:p></o:p>Rua Sílvio Lima, Pólo II, Universidade de Coimbra<o:p></o:p>3030-790 Coimbra<o:p></o:p>Tel: 239 796 217 (Ext: 421113)<o:p></o:p><o:p> </o:p></div></body></html>