In 2004, the US Department of Defense Advanced Research Projects Agency (DARPA) issued a Grand Challenge: the development of a driverless vehicle capable of covering a 150 mile test route, without human intervention. The best performing vehicle managed to travel less than eight miles. Eighteen months later, in October 2005, a second competition was run over a 132 mile off-road course. Five driverless vehicles completed the route successfully. In November 2007, the third competition challenged teams to produce a vehicle that could complete a 60 mile course in urban road traffic conditions, with the added requirement that vehicles obeyed California state driving laws. Vehicles from six teams completed the course within the six hour time limit. Within another three years, in 2010, Google was test driving self-driving cars on public roads. And today, an increasing number of US states have adopted, or are adopting, legislation that permits the use of autonomous vehicles on public highways.
In this two-part series of articles, we'll look at how robot technologies already have made significant in-roads onto our actual roads. We'll also take a peak at what the future may hold by way of some of the prototypes currently being demonstrated by the research and development departments of the major car manufacturers. But let's start with something that challenges many drivers - parking.
Ultrasound isn't just used for car reversing detectors, as this example from a free OpenLearn course describes in the context of ultrasound imaging in clinical settings.
If you're in the fortunate position of being able to afford to buy a new car, your car dealer will undoubtedly be keen to sell you a range of upgrades and optional extras. Forty years ago (in the 1970s), this might have included in-car audio systems (aka a car radio); thirty years ago (the 1980s), car alarms; air bags started to make an appearance twenty years ago (1990s), and GPS over the last decade (2000s). Over time, these optional extras start to become commodified, ultimately being "fitted as standard". So what sort of optional extra might we invest in today, or what technologies might even be leapfrogging the 'optional' part and making it straight into every production car?
Something that I think I'd certainly benefit from if I were to buy a new car would be support for car parking. In the simplest case, this might include a rear pointing camera, or maybe an ultrasonic distance detector to let me know how close the back of my car is to a wall, lamp-post or a parked car behind!
Even more convenient would be automated car parking, something that would allow the car to parallel park itself for me. A wide variety of manufacturers are now starting to offer such a facility, going by names such as "Active Park Assist" or "Advanced Parking Guidance". Here's an example, a demonstration of a Ford Focus parking itself:
In each of these cases, while the car is taking care of its own steering, the driver is still in control of selecting the gear and operating the brake and accelerator pedals. To a certain extent, we might think of the parking assist technology as a hands-on driving instructor who manages the steering for us while we operate the pedals. The vehicle's autonomy; that is, the extent to which the vehicle is able to operate under its own control, is limited. At any time, the driver "shall at all times be able to control his vehicle", for example by starting or stopping the car, and "shall in all circumstances have his vehicle under control so as to be able to exercise due and proper care and to be at all times in a position to perform all manoeuvres required of him" in accordance with sections 8.5 and 13.1 respectively of the Vienna Convention on Road Traffic, 1968. In passing, it's maybe worth noting that United States, for example, is not a signatory to this convention, and therefore not subject to the international rules it contains...
In many towns and cities, the space available for car parking is limited. Driver assist technologies may improve the efficient allocation of car parking space, with parking assist enabled cars using spaces that we might feel less than comfortable about reversing into. But what if we could make even more efficient use of space in multi-storey car parks, for example? Maybe we can...
In the OpenLearn article A dark future for warehousing?, I reviewed several robotics systems that are used to automate warehouse storage and retrieval. Some of these systems make use of a three-dimensional grid within which warehoused items are automatically placed and retrieved using something akin to a grid system of lifts and trolleys. What I hadn't realised was that a very similar idea is used in automated car parking facilities. For example, consider the Wöhr Multiparker in Budapest. As this video clip shows, once you've dropped off your car, the machinery takes over and parks it for you in whatever space it can find, although you won't necessarily know where it's parked!
Reassuringly, if you lose the token that will identify your car for retrieval purposes, the system is able to locate it from its number plate using ANPR - Automated Number Plate Recognition technology.
It's also possible to find footage of what automated storage looks like from the car's perspective.
Self-driving cars that park themselves
Given the very idea of automated car parking systems, can you think of other ways in which such a system might be implemented? This review of the Simmatec Automated Car Parking System describes several possible alternatives.
With parking assist, cars are already taking on some of the responsibilities involved with driving, albeit with the ability for the car to actually move under our direct control. When we're parking using an automated parking system, the space we're trying to park in is unlikely to change, at least while we're actually trying to park. Within an automated parking system, no driving is required - automated machinery is responsible for moving the car to and from its parking location. Driving on public highways, however, is another matter altogether, not least because of the uncertainty in the environment caused by the motion of other cars and pedestrians.
Even so, there may be certain situations where we can control a larger environment, not quite the public highway, but a constrained version of it. The aim would be to provide a simplified environment within which a driverless vehicle may be able to operate with reduced concerns for the safety of others: a multi-storey car park, for example. Retrofitting a multi-storey car park with an automated parking system is likely to be prohibitively expensive in the majority of cases, if indeed it is possible; but what if a car could be dropped off at the entrance to a multi-storey car park, and then left to park itself?
A demonstration by Audi at CES, the Consumer Electronics Show, in February 2013 demonstrated exactly this possibility. Here's a video of it in action (The technology behind Audi piloted parking):
Something of the technology behind the system appears to be described in this research paper by academics at the Institut für Neuroinformatik, Ruhr-Universität Bochum, the Institut für Informatik VI, Technical University Munich and representatives of Audi: Towards autonomous driving in a parking garage: Vehicle localization and tracking using environment-embedded LIDAR sensors. In particular, it describes how multiple LIDAR (Light detection and ranging) sensors are located in the garage in order to estimate the position and orientation of the car. This information can then be used to help guide the vehicle to an empty space.
This combination of environment and actor - car park and car - working together allows each part of the system to manage the part of the operation that it does best. For the car park, organising where there are spaces to park each car; for the car itself, actually getting into the parking space. While this approach may appear to offer the worst of both worlds, in that both car and garage need customising, the combined problem - of parking a car in an arbitrary space - is simplified.
In the next part of this series, we'll look at the extent to which today's cars are starting to take control of themselves while driving along on the public highway. We'll also be asking what challenges are still to be met in turning R&D activity into the reality of commercial production.
If you would like to learn more about robotics, you may consider studying the first-level course Technologies in practice from The Open University.