Dr. Haneen Alsehli is a Postdoctoral Research Associate in Professor Ivana Barbaric’s group at the Centre for Stem Cell Biology, University of Sheffield. She completed her PhD in Stem Cells and Regenerative Medicine at King’s College London, where she investigated how biochemical cues and physical confinement provided by synthetic hydrogel influence morphogenesis and differentiation in 3D human pluripotent stem cell models using high-throughput imaging. Her current research explores the mechanisms driving the selective advantage of genetically variant human pluripotent stem cells, aiming to develop strategies that enhance the safety and reliability of cell-based therapies.

The Role of Mechanosensing in the Selective Advantage of genetically variant human pluripotent stem cells

Abstract: 

Genetic stability of human pluripotent stem cells (hPSC) is of paramount importance for their use in regenerative medicine, but it is now well documented that hPSC can acquire genetic changes during culture. Notably, certain aberrations occur frequently, such as gains of chromosomes 12 and 17, and abnormalities in chromosomes 1, 20, and X. The recurrent nature of these aberrations suggests that variant cells harbouring them possess a selective growth advantage over wild-type counterparts. Indeed, our previous work demonstrated that variant cells have the ability to outcompete wild-type cells and eliminate them in a mosaic culture. Mechanistically, we showed that this was due to YAP-mediated mechanical cell competition. To identify the upstream regulators of YAP localisation, we performed transcription analyses of wild-type and variant cells and identified Caveolin-1 (CAV-1) as a putative candidate due to its upregulation in a range of variants compared to wild-type hPSCs. Next, we performed functional assays to explore the impact of CAV-1 on wild- type and variant hPSCs. Strikingly, we found that knockdown of CAV-1 reduces its expression in both Wild-type
and variant cells, therefore, strikingly regulates YAP nuclear/cytoplasmic localisation in variant hPSC. These findings suggest that Cav1 can be an upstream positive regulator of YAP, which could influence the selective advantage of variants hPSC overtake in co-culture. This approach will enable us to determine the relative role of mechanosensing at cell-ECM adhesions in the selective advantage of variant hPSC and devise strategies to suppress their occurrence.

Dr. Mohammed Basheer Mohiuddin is a postdoctoral researcher within the Interdisciplinary Research Center for Aviation & Space Exploration (IRC-ASE) at King Fahd University of Petroleum and Minerals (KFUPM) where he focuses on aerial robotics, reinforcement learning, embodied artificial intelligence, and autonomous systems. His research explores innovative solutions to real-world challenges in robotics, particularly in unmanned aerial vehicles (UAVs), by integrating machine learning techniques to enhance autonomy, intelligence, and adaptability. He aims to bridge the gap between theoretical advancements and practical applications, contributing to the next generation of intelligent robotic systems.
He completed his PhD in Robotics at Khalifa University, UAE, in 2024, where his research specialized in leveraging reinforcement learning to control robotic systems, with a particular emphasis on improving UAV autonomy and intelligence. He was a visiting research fellow at the School of Engineering and Technology, University of New South Wales (UNSW), Canberra, Australia, in 2024, where he collaborated on cutting-edge projects in robotics and autonomous systems. He earned his Bachelor of Engineering (B.E.) in Mechanical Engineering from Osmania University, India, in 2017, graduating as the university's top-ranked student. He then pursued his Master of Science (M.Sc.) in Aerospace Engineering at King Fahd University of Petroleum and Minerals (KFUPM), Saudi Arabia, which he completed in 2020.

Toward Trustworthy Learning-Based Control in Cyber-Physical Systems

Abstract:

The integration of learning-based methods such as Deep Reinforcement Learning (DRL) into Cyber-Physical Systems (CPS) promises unprecedented autonomy, adaptability, and control in complex environments. However, as these systems move from lab-scale experiments to safety-critical real-world deployments, one critical bottleneck persists: the lack of post-training closed-loop stability guarantees. This talk will explore a principled methodology for analyzing the stability of nonlinear dynamic systems controlled by trained DRL agents, using Lyapunov theory and linear-quadratic approximations to rigorously define the domain of attraction and stability margins.
Drawing from recent experimental work on a tower crane platform; a representative CPS with rich nonlinear dynamics, we demonstrate how DRL agents, trained entirely in simulation, can be deployed directly onto hardware with minimal fine-tuning. We present methods to quantify robustness to physical parameter variations (e.g., cable length, payload mass) and introduce a framework for Sim2Real transfer rooted in both classical control and modern learning paradigms.
This talk bridges the gap between autonomous learning controllers and verifiable CPS safety, proposing a pathway for trustworthy DRL deployment in industrial applications such as smart manufacturing, robotics, and infrastructure systems.