Let’s Talk: Dr Graeme Smith on Mobile Autonomy
Dr Graeme Smith is a recognised expert in Advanced Vehicle Control systems and the Connected Car, and holds several patents in the areas of Adaptive Cruise Control and the application of Impulse Radar systems to vehicles. Graeme has held executive leadership positions in major car companies as well as several global start-ups and Joint Ventures, building and leading international engineering teams in electronic/software product development, service delivery and customer support.
What is Oxbotica?
Oxbotica was spun-out from Oxford University’s internationally acclaimed Mobile Robotics Group (MRG) in 2014. Our founders, Professors Paul Newman and Ingmar Posner, are Principal Investigators with MRG, which was formed to “invent the technologies that allow machines to ask and answer ‘Where am I?’, ‘What surrounds me?’ and ‘What should I do next?’”. So we’re reaping the benefits of 130 man-years of leading edge research in the area of mobile autonomy, and developing applications that allow robots, vehicles, machinery and people to precisely map, navigate and actively interact with their surroundings.
What are the key components of mobile autonomous systems?
We start by developing a detailed 3D map of the target environment. Then we add in the ability to understand in real-time what’s around us, whatever environment we are operating in. That could be inside a warehouse, in the home, in a mine or whilst driving on the road where we have to recognize, for example, other vehicles, racking, pedestrians, and static objects like railway lines, trees or buildings. And then putting it all together so the machine – a vehicle or a robot – knows where it is and what’s around it, and can safely and efficiently navigate to where it wants to go.
What’s special about our approach is our unique spectrum of integrated technologies, including our ability to create highly accurate 3D maps of the target environment quickly, cheaply and robustly.
With our patented approach using lasers and vision sensors (cameras), we can map environments simply by moving through them.
These maps in turn enable machine localization as well as scene perception and navigation – where am I, what’s around me, how can I get from here to there safely and efficiently. And in addition to self-driving vehicles, 3D maps are the foundation blocks for multiple applications ranging from Smart City planning, surveying, infrastructure management, asset detection/management and low cost localisation for devices ranging from smartphones to cars, buses and trains.
What are some of the projects Oxbotica is working on?
One of the most interesting and ambitious projects Oxbotica is involved in is the GATEway project in Greenwich, London. This project, funded by the UK government with matching funds from consortium members, aims to “establish the UK as a global hub for the development of autonomous vehicle technologies and to integrate driverless vehicles into existing urban environments by trialling them in two major UK cities.”
A key element of the project is the development of a number of autonomous shuttles in conjunction with our partners at Heathrow airport. The project will result in around eight autonomous shuttles in public operation for a six-month period starting in July 2016. This project should provide valuable lessons and experience on the opportunities and challenges of using autonomous shuttles as an alternative form of public transport.
Oxbotica’s role in this effort is to deploy our Selenium autonomous control systems and sensor sub-systems that enable the operation of the self-driving pods in different types of settings around Greenwich. We are also providing a novel shuttle management service that integrates scheduling and co-ordination of the shuttles, as well as data sharing within our new cloud-based Caesium fleet management system.
What are other applications of mobile autonomy?
There are many. A common application is warehouse automation. A forklift truck is relatively inexpensive; most of the operational cost is the cost of the labor. Using robots in a warehouse to autonomously move goods can quickly lower costs while improving productivity and safety. Automation also enables warehouse operators to scale up or down very readily.
There is also a lot of interest at the moment in robots that can go inside an industrial or petrochemical facility to do inspections and maintenance. In many instances, a facility might have to shut down and perhaps be decontaminated before a technician can safely conduct an inspection or perform maintenance. So a very robustly designed robot that can operate effectively in toxic or hazardous environments offers a number of advantages.
Autonomous transport systems in closed environments like a retirement community or university campus are another example. In closed environments, it’s possible to greatly facilitate mobility in a very safe, flexible and user-friendly manner by using autonomous vehicles to connect different places within the campus. And since the autonomous vehicles don’t have to share the road with “regular” vehicles – as they will, for example, in Greenwich – that gives us a lot more flexibility. We can relax some of the expensive crash protection and design a lighter, more delicate structure. The sensors and control systems also don’t need to be as robust.
Autonomous devices, connected objects, the Internet of Things … there is a tremendous amount of activity taking place these days in these areas. Why now?
While several things have occurred to make different applications feasible, the most important reason is that a supercomputer isn’t needed any more to do many things. We can do some very complex and sophisticated applications today with fairly affordable electronics.
Sensor costs, for example, have come down dramatically. Ten years ago, the sensors on an autonomous prototype would have cost 10 to 100 times more than today. Lasers used in autonomy to tell the distance to something are also coming down in price. Cameras that help us understand the angles between things have come down enormously in price; we can find cameras now for a couple of dollars.
And the ability of software to use what these devices are “seeing” has increased dramatically. Considerable research has been done in this area that’s now starting to generate results. So we now have hardware and software that can accurately identify people and objects, like other cars, very, very quickly.
So it’s partly the costs of these technologies improving, it’s partly the availability of very fast, low cost processors with cheap memory, and it’s partly the research that’s been done on how to design and engineer different applications.
What are the aims of the collaboration between Oxbotica and XL Catlin?
We’re very excited about teaming-up with XL Catlin and combining our know-how in mobile autonomy with XL Catlin’s deep expertise in risk management, insurance and data management across many industries.
We’re engineers and scientists. While we can envision how different applications could change the risk landscape, we certainly don’t have the depth of understanding about risk, and especially liability risk, that XL Catlin has. There are also important questions and implications about data in the future: how it is captured, managed and used in multiple domains, for example, to improve future autonomy systems in order to predict outcomes and, in turn, create different risk and management models.
The applications we’re developing – self-driving pods, driverless cars, robots – all transfer at least partial control of a machine from a person to an automated system. And since people are fallible, mobile autonomy solutions are typically safer and more reliable. The self-driving pods at Heathrow Airport, for example, connect the perimeter car park and Terminal 5, a 3.5km journey; since the system opened in 2011, they have carried 1.5 million passengers without a single accident.
Nonetheless, understanding and evaluating how the use of autonomous systems impacts existing risks and creates new risks is critical.
As we pave new ground with different technologies, we’re also raising new scenarios when it comes to considering risk and liability. By working together, XL Catlin will be right at the interface between technology and society, extending its deep expertise in evaluating, modelling and managing risk. Oxbotica, in turn, will benefit greatly from XL Catlin’s know-how and capabilities throughout every stage of the process – from design to testing to development to ongoing management.
Last question. C-3PO or R2-D2?
R2-D2 – we love robots with wheels!