Computer Science and Engineering | University at Buffalo’s Post

– My research could potentially help mapping the seafloor more efficiently and with higher resolution which would help scientists understand the ocean better, says Yiping Xie who is defending his doctoral thesis today at 14:00. Yiping Xie, WASP PhD student at the Division of Robotics, Perception and Learning at KTH Royal Institute of Technology, is defending the doctoral thesis “Bathymetric Surveying Through Neural Inverse Sonar Modeling” today, June 5th, at 14:00. Join the public defense online or in E3, Osquars backe 14, Stockholm. More information and link to thesis 👉 https://lnkd.in/dGEGbfpY What are the main findings of your doctoral thesis? – Inspired by 3D reconstruction from camera images, we can use similar deep learning based approaches to reconstruct bathymetry from imaging sonar data. In what way can your research be of importance to our society in the future? – My research could potentially help mapping the seafloor more efficiently and with higher resolution which would help scientists understand the ocean better. Main supervisor: John Folkesson #wasp #waspgraduateschool #phd

Yesterday's seminar at the QUT Centre for Robotics left us enlightened, thanks to the prolific presenter, Richard Hartley, a trailblazer in Computer Vision and a Distinguished Professor Emeritus at the The Australian National University. The session, titled "Diffusion and Text-Based Guidance for Image Morphing," delved into the realm of generative models, specifically diffusion, and its application in estimating probability distributions for natural images. Hartley's exploration of paths through high probability regions of image space unveiled a method for creating seamless transitions between different images, offering a glimpse into the future of image morphing. With a rich background in mathematics and computer science, Hartley's illustrious career spans academia and industry, including impactful stints at GE's Research and Development Centre and Google Research, Switzerland. His pioneering work in projective geometry techniques for computer vision, showcased in the seminal book "Multiple View Geometry in Computer Vision," has left an indelible mark on the field. 🚀👁️ #ComputerVision #Robotics #TechInnovation

A team led by computer science Ph.D. student BOTAO HE invented a new camera mechanism that improves how robots see and react. Inspired by the human eye, the innovative camera system mimics the eye's tiny involuntary movements to maintain clear and stable vision over time. The team's prototyping and testing of the camera (called the Artificial Microsaccade-Enhanced Event Camera, or AMI-EV) was detailed in a paper published in Science Robotics. The paper was co-authored by computer science faculty members Yiannis Aloimonos and Cornelia Fermuller, along with Jingxi Chen (B.S. ’20, computer science; M.S. ’22, computer science) and Chahat Deep Singh (M.E. ’18, robotics; Ph.D. ’23, computer science). Read more: https://go.umd.edu/camera #computervision #computerscience #phdstudent #phdlife #camera #prototype #published #scienceterp #umd UMIACS, Department of Computer Science, UMD

The University of Texas at Austin Learning Agents Research Group, Department of Computer Science is working on LIBERO, a lifelong robot learning benchmark, lifelong learning in decision making (LLDM) to study knowledge transfer in decision-making and robotics at scale. LIBERO paves the way for prototyping algorithms that allow robots to continually learn. LIBERO, a benchmark that is tailored to the nature of lifelong robot learning — a benchmark that can continually grow. UT Austin build this benchmark to invite communities of machine learning and robotics people to study a core problem in lifelong learning: knowledge transfer. The LIBERO Team: Peter Stone, Bo Liu, Yifeng Zhu, Chongkai Gao, Yihao Feng, Qiang Liu, Yuke Zhu https://lnkd.in/gCj-dD4Y

The Rutgers University Computer Science Department proudly announces that Professor Kostas Bekris, together with UMass Lowell (Lead), Univ. of South Florida, Worcester Polytechnic Institute and Yale University, have been awarded a $1.5 million grant under NSF’s Pathways to Enable Open-Source Ecosystems (POSE) program for their project entitled “NSF POSE: Phase II: Collaborative Open-source Manipulation Performance Assessment for Robotics Enhancement (COMPARE) Ecosystem.” This project is funded by the Pathways to Enable Open-Source Ecosystems (POSE) Program which seeks to harness the power of open-source development for the creation of new technology solutions to problems of national and societal importance. COMPARE ecosystem will enable the robot manipulation community to more effectively conduct research and evaluate system performance, with the goal of enabling the quantitative comparison of approaches. By establishing a new, community-driven, open-source ecosystem for robot manipulation research and benchmarking, the field will move more rapidly towards solutions to the open problems of robot perception and grasping. This is a crucial step to push the field forward as the ineffectiveness of these capabilities limits progress in robot manipulation. Developing the COMPARE ecosystem will produce the first systematic community-driven framework to roadmap the current open-source landscape, develop mechanisms and strategies to improve the field, and establish activities that will actualize these improvements, resulting in a more effective and harmonious ecosystem for robot manipulation. To learn more about the POSE grant visit:  https://lnkd.in/e7pfmckQ

🚀 I am happy to announce that our team will present four cutting-edge research papers at the upcoming #ICRA conference, one of the leading conferences in the field of robotics! This would not have been possible without the hard work of the students I am working with, and the collaborations with other colleagues from the The University of Edinburgh, and UCL (University College London). Here is the list of our accepted papers: 1) Lifelong Robot Library Learning: Bootstrapping Composable and Generalizable Skills for Embodied Control with Language Models  Georgios Tziafas, Hamidreza Kasaei 2) Harnessing the Synergy between Pushing, Grasping, and Throwing to Enhance Object Manipulation in Cluttered Scenarios   Hamidreza Kasaei*, Mohammadreza Kasaei* (*equal contribution) 3) Self-supervised Learning for Joint Pushing and Grasping Policies in Highly Cluttered Environments  Yongliang Wang*, Kamal Mokhtar*, Cock J. M. Heemskerk, Hamidreza Kasaei (*equal contribution) 4) TiV-ODE: A Neural ODE-based Approach for Controllable Video Generation From Text-Image Pairs  Yucheng Xu, Nanbo LI, Arushi Goel, Zonghai Yao, Zijian Guo, Hamidreza Kasaei, Mohammadreza Kasaei, Zhibin (Alex) LI 📌Stay tuned! The arxiv version of the papers and their supplementary materials will be released soon. 

With a National Science Foundation (NSF) grant, Asst. Prof. Xian Yu and colleagues at University of Michigan and Arizona State University will work to enhance safety of reinforcement learning, a feedback-based #machinelearning technique 🤖💻 https://lnkd.in/du9RMT3z “This is a collaborative project that spans multiple disciplines, including operations research, computer science, manufacturing and mechanical engineering,” said Yu. “In this project, we aim to understand the fundamental properties of the risk-sensitive distributional reinforcement learning (DRL) framework.”  The funding is part of a $10.9 million investment by the NSF to support research that will help ensure advances in artificial intelligence (AI) go hand in hand with user safety. The objective of the Safe Learning-Enabled Systems program, a partnership between NSF, Open Philanthropy and Good Ventures, is to foster foundational research that leads to the design and implementation of safe learning-enabled systems — including autonomous and generative AI technologies — that are both safe and resilient.

Excited to share I have started my PhD at the University of Surrey. My research will focus on Robotic Analogy Transfer, exploring how machine learning algorithms can transfer previously learned skills to new scenarios. #PhD #Research #MachineLearning #UniversityOfSurrey

TOWARDS ROBOTICS ECOSYSTEMS: FEASIBILITY AND ACCEPTABILITY Humanity is getting better and better at building embodied (robots) and abstract (algorithms) agents which are able to perform complex tasks. These agents reached such a degree of complexity that we are not always able to explain. Moreover, they are becoming more and more integrated within human society. Are we creating a new species? A new form of (artificial) life? A robotic ecosystem? On 23 september, top scientists will discuss the actual feasibility of this perspective from the technical point of view and the possible reception of it by humans. Join us at Trieste Next, with: 🎙 Ruth Aylett, professor of Computer science Heriot-Watt University in Edinburgh, writer and poet 🎙 Gusz Eiben, professor of Artificial Intelligence Vrije Universiteit Amsterdam (VU Amsterdam) 🎙 Eric Medvet, professor of Computer engineering University of Trieste 🎙 Erica Salvato, researcher in Automation University of Trieste Info ► u.garr.it/cDzBi