This November, researchers at the Oxford Robotics Institute (ORI) are sharing their work at a conference dedicated to improving mobile autonomy. ORI have seven members attending the 21st IEEE International Conference on Intelligent Transportation Systems (ITSC), two of them are leading a special session titled “Beyond Traditional Sensing for Intelligent Transportation”.
This session is motivated by the gap between sensor technology and its use. Due to the recent surge in robotics, sensors have become more advanced and span an increasing number of modalities. Despite the improved capabilities and breadth of available sensor systems, those used for intelligent transportation have remained relatively uniform across platforms. A typical sensor suite features camera, lidar, GPS, and inertial sensors.
As a consequence, algorithms and techniques designed for mobile autonomy do not take full advantage of the rich information offered by modern sensors. Since all tasks, including perception, localisation, decision-making, and learning, are built on top of sensing, ORI believes that exploring alternative approaches to sensing is a compelling research area that would improve the overall robustness and accuracy of transportation systems.
ORI aims for the special session to explore unconventional sensing for intelligent transportation in three ways. The first objective is to investigate sensor systems that are not typically applied to certain transportation tasks, such as radar for precise localisation and audio for failure detection. The second objective is to consider untraditional sensor configurations and placements, such as ground-facing cameras using shadows to detect occluded moving objects. The third objective is to look into commonly overlooked sensing information, such as the use of atmospheric sensors for gauging road surface traction or in-vehicle sensors for driving analysis.
Large-scale mobile autonomy, both indoors and outdoors, represents one of the major research areas of interest of ORI. Under the EPSRC Program Grant titled “Mobile Autonomy: Enabling a Pervasive Technology of the Future” that was awarded in 2015, the Institute has been investigating various research topics that would enable large-scale mobile autonomy, from navigation and mapping to scene interpretation and learning. Our research provides solutions for a variety of application domains, including infrastructure inspection, space exploration, off-roading, and transportation.
ORI aims to build autonomous systems that not only operate safely and robustly in complex and time-changing environments, but also improve their performance over time. ORI strongly believes that robust, reliable, and fully informative sensing plays a key role towards the realisation of this vision, and the exploration of untraditional perceptual schemes is a now an essential step towards the development of such systems.
This special session aims to bring together researchers interested in transportation and sensing to encourage synergies across the two areas by stimulating discussion on the advantages of unconventional sensing. It will contain research presentations on novel work and feature a keynote talk by Professor Tim Barfoot from the Autonomous Space Robotics Lab at the University of Toronto. It is chaired by ORI researchers Dr. Letizia Marchegiani, who works on auditory perception and machine listening, and Sarah Huiyi Cen, who works radar-only ego-motion estimation and localization. ORI hopes that many will join us for what will be an important moment of dialogue and analysis in the areas of intelligent transportation and, more generally, autonomous systems.
Other members of ORI whose submissions will be presented at ITSC include Prof. Paul Newman, Dr. Lars Kunze, Tom Bruls, Horia Porav, and Tarlan Suleymanov.