Watch Again | IEEE UK and Ireland Robotics and Automation Society Webinar – FMCW Radar, What’s Next? From Off-Road to Doppler by Daniele De Martini, Oxford University
Abstract
Frequency Modulated Continuous Wave (FMCW) Radar has emerged as a robust and reliable sensing modality for navigation tasks, owing to its long-range sensing capabilities and resilience to adverse weather and lighting conditions. The Mobile Robotics Group (MRG) has been at the forefront of leveraging this sensor technology for various navigation tasks, including odometry, localization, object detection, and planning, over the past years. However, much of the research has been confined to urban environments, where the advantages of radar are less crucial due to the abundance of environmental features.
In this talk, after a brief overview of MRG’s prior work, De Martini will introduce their newest dataset, SAX. This dataset encompasses a comprehensive range of environments and scenarios that an autonomous vehicle might encounter, from bustling avenues in London to snowstorms in the Scottish highlands. He will present early results demonstrating the effectiveness of radar-based localization in these challenging scenarios.
Furthermore, he will explore a new functionality of the Navtech sensor, which enables the detection of velocities in the scene through the Doppler effect, manifested as ripples in the raw spectrum of the sensor return, showcasing how this capability can enhance odometry estimation, underscoring the potential of this new modality in autonomous navigation.
About the Speaker
Daniele De Martini is a Departmental Lecturer in Mobile Robotics at the Oxford Robotics Institute, where he co-leads the Mobile Robotics Group. He is also a College Lecturer in Engineering Science at Pembroke College.
Daniele’s research focuses on creating intelligent, autonomous robots that can work with or for humans, mainly by applying Artificial Intelligence (AI) techniques. He has worked on navigation and scene understanding, including mapping and localisation, detection, and segmentation, using different sensing technologies – from vision to laser to radar – and in diverse environments and conditions – from central Oxford to snowy Highlands.
Another prominent research interest is in the interplay of robotics and smart infrastructures to allow sensing and computing to be shared and elastically allocated for safe operation.