Congratulations (Dr.) Russell Buchanan for successfully passing his DPhil viva

Title: Inertial Learning and Haptics for Legged Robot State Estimation in Visually Challenging Environments

Congratulations to Russell Buchanan for successfully passing his DPhil viva on 9th June 2023. Russell completed his DPhil at the University of Oxford and joined the University of Edinburgh as a PDRA in April.

His thesis is entitled ' Inertial Learning and Haptics for Legged Robot State Estimation in Visually Challenging Environments.

Summary:

This thesis focuses on addressing the challenges faced by legged robots in terms of accurate state estimation, particularly when visual sensors fail in visually challenging conditions like darkness or smoke. Legged robots have great potential for automating dangerous or dirty tasks, as they can navigate through difficult terrains such as stairs or mud. However, the lack of robust localisation has hindered their widespread deployment.

First, haptic sensors are used to compare geometric information with a prior map, enabling a robot to localise without vision. This is similar to a blindfolded person in their own bedroom. They can reach out to touch their furniture and immediately localise. Next, haptic sensors are used to classify different terrain types (like grass or concrete) as additional information for localisation. Finally, new techniques in deep learning for inertial sensors used to improve state estimation.

The deployment of legged robots has the potential to benefit society by automating dangerous, dull, or dirty jobs, as well as assisting first responders in emergency situations. This thesis contributes to the field by addressing the challenges of state estimation and paving the way for the real-world deployment of legged robots in various applications.

Examiners:

Professor Niki Trigoni (University of Oxford, Internal)

Professor Dimitrios Kanoulas (University College London, External)