Learning to see latent variables: Robotic state estimation made scalable. This project aims to develop a novel optimisation-based framework to ensure computationally efficient and resilient real-time

Description

Learning to see latent variables: Robotic state estimation made scalable. This project aims to develop a novel optimisation-based framework to ensure computationally efficient and resilient real-time estimation of latent variables. Robots have numerous unmeasurable latent states crucial for decision-making, monitoring, prediction, and for designing controllers that interact with the real world. However, challenges in computational scalability and long-term performance in current estimation methods are not well understood. This research will lead to new knowledge, approaches, and algorithms that achieve high-performance robotic estimation. Such advancements will benefit robotics, industrial automation, control engineering, and other fields that demand state estimation within the broader Australian communities.. Scheme: Discovery Early Career Researcher Award. Field: 4007 - Control Engineering, Mechatronics and Robotics. Lead: Asst Prof Bowen Yi