A radar-based approach to localisation in mobile robotics

Lead Research Organisation: University of Oxford
Department Name: Engineering Science

Abstract

The Oxford Robotics Institute creates, runs, and exploits the world's leading research programme in mobile autonomy, addressing fundamental technical issues which impede large-scale commercial and societal adoption of mobile robotics.

We need to build better robots - we need them to be cheap, work synergistically with people in large, complex and time-changing environments, and do so for long periods of time. Moreover, it is essential that they are safe and trusted. We are compelled as researchers to produce the foundational technologies that will see robots work in economically and socially important domains. These motivations drive the science in this DPhil.

For a widespread adoption of autonomous vehicles to be feasible, the technology should be fully functional without depending on specialised infrastructure. This self-reliance ensures scalability and allows for implementation in many different environments, despite limited access to external support like roadside infrastructure or GPS. This technology is set to reduce road accidents, provide increased mobility to those who cannot drive, and minimise traffic congestion through more efficient driving techniques. Autonomous systems rely on a variety of sensors to perceive their surroundings that can be fallible in certain conditions like bad weather, where any predictions made could make autonomy more challenging than usual. By implementing a form of introspection, poor predictions can be recognised and the associated uncertainty handled accordingly. Knowing when to have less confidence in observations is important for reliable autonomy in adverse circumstances.

I will research the algorithms that will allow autonomous vehicles to operate in challenging environments using radar technology. While lasers and cameras work well in clear conditions, localisation and perception tasks can run into difficulty in fog or poorly lit environments. Radar is able to relay detailed information about a robot's surroundings, but interpreting that information presents difficulties not encountered in vision-based approaches. Frequency Modulated Continuous Wave (FMCW) radars can be used to perform 360-degree scans of an environment and return information at much greater rangers than other sensors, but introduce some unique challenges. Multipath reflections and sidelobe radiation can introduce ghost objects in empty spaces. These factors lead to uncertainties which must be taken into account in any proposed solution that relies on radar.

Besides the engineering challenges presented by autonomous vehicles, one also has to consider the cost of the sensors and computer hardware. In order to achieve widespread adoption of autonomous software, the technology needs to be able to run on low-cost hardware using low-cost sensors. This is particularly important in the context of radar where cost can quickly make a solution unfeasible. This opens up many new questions on how to achieve this in a cost-effective manner as mobile autonomy transitions from the research environment into the real world.

Prof Paul Newman is supervising this research. This project falls within the EPSRC Engineering research area.

Publications

10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/R512333/1 01/10/2017 30/09/2021
2261371 Studentship EP/R512333/1 01/10/2017 30/09/2021 Roberto Aldera