Abstract: Human movement is complex and highly variable, presenting major challenges when designing assistive technologies that must work reliably in real-world settings. In this talk, we will discuss how assistive robotics, exoskeletons and functional electrical stimulation (FES) can be used to personalise rehabilitation and support human motion. I will highlight recent work in gait training, human-in-the-loop optimisation and adaptive human-robot-FES collaboration, emphasising the need for individualised interventions that respond to unpredictable human dynamics.

Abstract: Recent machine learning (ML) research has chiefly focused on
optimising performance on tasks whose structure is fully specified in
advance. For instance, with assembly tasks, the agent has complete
knowledge prior to deployment about which component parts form valid
structures; e.g., that fire trucks must have ladders. The agent may
not know, and will have to learn, what fire trucks or what ladders
look like, but they have a complete and accurate hypothesis space of

Abstract: This is going to be an educational lecture, rather than detailing novel research. I am going to look under the hood of the fundamental operation of linear optimization that we all do under different umbrellas. Specifically, I am going to discuss the geometric and algebraic subtleties of solving linear systems, least squares, and, in more generality, variational and weak solutions (and why we call them this way). The purpose is to promote a deeper understanding of how optimization functions in its most basic level, which is something that is usually overlooked.

Abstract: In this talk, I will describe the work done in the context of the Japanese Science and Technology (JST) Moonshot projects that we have been involved in since the past year.

Abstract: This talk introduces SoftManiSim, a fast, open-source simulator for multi-segment continuum robots that models dynamic loading and contact. Designed for real-time use and deep learning integration, SoftManiSim supports parallel simulations and generates high-quality synthetic data for training. I will highlight two key applications: (1) using the simulator to develop a framework for shape-aware control of tendon-driven continuum robots; and (2) applying the simulation framework in a design study aiming to enhance continuum robot stability and safety.

Abstract: our recent work proposes a robot system which contains three main components: 1. Data processing module for extracting human hand motion trajectory and object motions. 2. Retargeting module based on MANO hand model for mapping from human motions to robot motions. 3. Residual action learning with RL for learning how to refine imperfect motion priors. With the proposed system, robot can quickly learn how to complete complex dexterous manipulation task with one single human video demo. 

Abstract: Human feedback is widely considered the 'gold standard' for AI alignment, but what if this 'gold' reflects inherently diverse and conflicting human views on what constitutes ‘good’ and ‘safe’? This keynote will explore the technical and ethical challenges posed by differing viewpoints and opinions across individuals and groups.

Abstract: our recent work proposes a robot system which contains three main components: 1. Data processing module for extracting human hand motion trajectory and object motions. 2. Retargeting module based on MANO hand model for mapping from human motions to robot motions. 3. Residual action learning with RL for learning how to refine imperfect motion priors. With the proposed system, robot can quickly learn how to complete complex dexterous manipulation task with one single human video demo. 

Talk on Generative AI in design support, In EM2.44  11AM-12PM on Thursday 19th June

Speaker: Dr. ANDREA GENEROSI PROFILE

Speaker: Zhaoxing Deng

Title: Automated Skill Evaluation of Bronchoscopy Operators Using Geometric Feature Analysis