Google has been in the automobile industry for a while now. The internet giant has dabbled with cars for almost a decade, but not until the recent years have its efforts been truly noticed, talked about and even had a major impact on the rest of the industry. In fact, Google is now addressed as one of the very first movers in a segment that is expected to lead automobile innovation in the coming years, and that is owing to this one particular crazy experiment.
The Google Self-Driving Car Project
Quite aptly referred to as ‘Project Chauffeur’, the Google Self-Driving Car Project begun way back in 2009, with only one goal in sight – building a completely autonomous car by 2020. The idea was to build one prototype that would be able to successfully drive itself across real-life scenarios, and then replicate the technology en masse.
How it all began
Google chose Stanford University professor Sebastian Thrun to head this project. Interestingly, Thrun was working on making cars drive themselves for a while, and had won the $2 million grand prize of the Defense Advanced Research Projects Agency (DARPA) Grand Challenge in 2005. He led a team of Stanford students and faculty to built Stanley – a ‘robot car’ based on a 2004 Volkswagen Touareg R5 TDI, fitted with an Intel-powered six-processor computing platform, and a host of actuators and sensors. Aiding the Stanley robot car across nearly 230 kilometres through the Mojave desert were experts from The Volkswagen of America, Intel Research and others.
It is this expertise, along with the basic formula of using peripheral imagery and feeding it to onboard computers to make a car drive itself, that Thrun brought to Google. In 2009, the Google Self-Driving Car Project began with other notable engineers like Chris Urmson (who won the same DARPA Challenge in 2007) and Anthony Levandowski, who has more recently been the centre of the Uber v. Google controversy.
Alongside Thrun, Levandowski has been one of the key first innovators in the autonomous vehicle space. He founded two companies that were later acquired by Google – 510 Systems and Anthony’s Robots. While the former worked on Lidar modules, which as we know now are crucial for self-driving cars, the latter was focused on getting an actual self-maneuvering car on public roads. Levandowski used a Toyota Prius, calling it ‘Pribot’, and fitted it with a Lidar unit and other technology that he himself was working on, and succeeded in putting out a trial on public terrain.
Google’s first self-driving car, the Toyota Prius
The Pribot, while not belonging to Google, can be hailed as the first indirect predecessor to Google’s eventual self-driving efforts. The company brought Thrun, Levandowski and 13 other engineers together to form a team, and got a fleet of six Toyota Prius and one Audi TT as the test rig. Along with Lidar modules, each of these vehicles included cameras, imaging sensors, an onboard computer circuit and necessary signal processors and algorithms that would automate customised levers to operate acceleration, braking and steering – the three key functions. The initial goal was to drive 10 separate 100-kilometre stretches without accidents and interruptions, in full autonomous mode.
Needless to say, this got the ball rolling more than successfully, and laid the foundation for other technology and automobile companies to follow.
Fun fact: It is this experiment that opened a new job designation at Google, that of a driver with perfect driving records. The company, now an independent entity Waymo parented by Alphabet, continues to hire for this designation even today.
On-road success, and Firefly
Beginning with the first Toyota Prius vehicles, Google’s on-road self-driving experiments gradually turned into success. This was a gradual process, and one that saw Google constantly evolve its technology. The company further added three Lexus RX450h crossovers to its fleet, cars which would soon become immediately identifiable with Google.
One big milestone was the licensing of the first autonomous vehicle in history, when one of Google’s self-drive Toyota Prius cars was registered in Nevada, where it was testing its cars. The pattern chosen for identifying self-driving cars was a red background, with the infinity symbol to the left of the number plate, denoting ‘one for the future’.
Simultaneously, it is Google’s efforts in the autonomous vehicle space that first sparked off the legal debates. While it primarily originated in USA, the context was to adjust rules pertaining to Artificial Intelligence taking over. All driving rules, pretty much all over the world, have required the compulsory presence of a human in sound mental and physical condition behind a steering wheel. Now, with the need for steering wheel in question and drivers set to become mere passengers, the lawmakers called for discussions to revise laws pertaining to autonomous cars.
Regional implementation of laws surrounding autonomous cars were first facilitated by Google, beginning with the state of Nevada. Other states in the USA that now permit autonomous vehicle testing on public roads include Florida, California, Michigan and Texas. The laws have, for now, made the presence of a human imperative in any self-driven car. The person behind the wheel must also be in sound condition, and be able to intervene into the systems and take charge of the autonomous car whenever necessary. As a result, the person is also required to remain alert to his surroundings.
Waymo’s flagbearer – the Chrysler Pacifica Hybrid
The on-road success has seen Google notch up over two million reported miles, making it the most experienced company so far when it comes to on-road autonomous driving. The company has since replaced its existing fleet of test cars with 23 Lexus SUVs, and more recently, struck a deal for 100 Chrysler Pacifica Hybrid minivans to test autonomous, ride-sharing journeys on road in Phoenix, Arizona.
Through all this, Google also built its own car, and killed it too. The Firefly was shown in its prototype form in May 2014, and it was unlike anything that we had seen before. It looked affable, was shaped like a pod, and only had two seats inside its cabin – no steering wheel, no pedals, nothing. It was launched by 2015, and was fitted with a beacon-like Velodyne 64 Lidar unit that provided a high definition, precise map of the surroundings to the car’s computers. That, along with the onboard cameras, sensors and computers, helped it gauge its way around without any possible human interference.
Firefly is particularly important for four reasons. One – it was the first car that Google really built for itself, in association with a few other companies of course. It stands as Google’s first car that was used to show the fruits of its alleged $1.1 billion investment in the segment till now, over eight years. Two – it not only became the face of Google’s self-driving efforts, but of all autonomous car efforts worldwide. Three – it gave the world’s first truly autonomous ride on public roads to a common citizen, when Steve Mahan, a legally blind person, was taken around without any supervision at all. And finally, it is the short-lived Firefly on which Google spun off its self-driving car project to Waymo – a completely independent entity under its parent, Alphabet.
Waymo, and the way forward
With most of the brunt work of establishment achieved, Waymo’s present task is to teach the intricate bits to its cars. The company has initiated the ride-sharing model with its Pacifica minivans, has bid goodbye to the Firefly, which served as a noble prototype for testing, and is now engaged in making the cars practical outside of the pre-mapped, controlled roads.
Waymo’s objective is to make its cars viable across all roads, where they would be able to identify minor obstacles from major ones, realise manual traffic alerts a-la police officers, and iron out these glitches from the software. Of the 15 accidents that Google (and Waymo) have been involved in, the other human drivers were at fault for 14 out of the 15 times. For the solitary mistake that its cars have made, it mistook sandbags as obstacles that needed to be avoided, and veered off course to bump into a bus that was passing on its side lane.
Waymo will also intend to refine the design of these cars, and find a way to retrofit the Lidar and other sensors in a way as to not disrupt the visual appeal of our future vehicle. In short, Waymo now has all the hardware it needs for the autonomous cars. All it needs now, is the perfect software.
Android Auto
It is in search of this perfect software that Android Auto came into being. While self-driving cars are certainly much more exciting and challenging, the need to provide a familiar infotainment interface for the millions of Android smartphone users worldwide became a compelling case for Google.
Google’s most successful auto effort so far
At I/O 2014, Google announced Android Auto as a simplified, ergonomic interface for drivers to access connected applications via a car’s infotainment system. A few months earlier, at the Consumer Electronics Show 2014, Google announced the Open Automotive Alliance (OAA) – yet another consortium that would indicate the nature of collaborative alliances that have now become common in the automobile space.
The OAA is a group of companies that aim to use Android as the platform of operation within cars, not just for infotainment. This alliance suggests the use of Android OS as the core platform on which car systems can be built, and have their own suite of connectivity options and apps. This, in turn, will facilitate phone-free usage of car systems, which till now have been heavily reliant on phones. The OAA has Nvidia as the prime chip provider, and Google as the software maker, with a wide range of notable car makers and in-car system manufacturers listed as partners.
Nevertheless, the first leg of the OAA’s mission has been to establish Android Auto as a de facto infotainment standard in cars, and to an extent, it has managed to strike a balance between form and functionality, although not without compromises. First introduced with Android Lollipop (v5.0), Android Auto presents a familiar Android interface with floating cards that provide information, and can be operated via touch, physical control buttons on the car’s dash, or voice. It also integrates the Google Assistant voice assistant that allows users to feed navigational information, make calls, change songs and more. Application support has been Android Auto’s greatest strength, and along with Google Maps, Play Music, Android Auto supports a host of third party applications like WhatsApp, Spotify and more.
For a more detailed analysis of Android Auto, and how it fares against Apple CarPlay, click here.
Android in Cars
The next step, seeing that Android Auto is almost nearing maturity, is to present systems that run directly on Android OS. These systems will benefit from the rising connected car telematics, wherein in the car itself integrates data connectivity into its system, and the system functions pretty much like an independent smartphone.
Your future car will run on electricity… and Android
As a result, a user can leave their phone behind, and still get all call and message prompts on his car system, as long as the accounts are synced. Alongside, the system will integrate all car dynamics, mechanical data, sub-systems (air conditioning, seat adjustments, etc.) along with an app store, and live GPS data, to aid cross-car connectivity.
In functionality, Android in cars is similar to the likes of Automotive Grade Linux, which also aims to provide an open source platform for automobile manufacturers to collaborate and build on. Having a common software platform helps in tackling security threats, and also build application support. As of now, the likes of Volvo and Audi have partnered with Google in this space.
While Google has enjoyed a head start in the autonomous car space, and even with its system-on-interface car system Android Auto, it is seemingly yet another contestant in the connected car race. Despite being one of the most successful with its prototype tests, Google is still tangibly far from introducing a driverless car running on its technology. All of this brings up one point, where Google, despite practically being the forefather of sorts for connected car technology, has not yet managed to deliver sound end products yet (apart from Android Auto, which also has many areas to improve on).
The next three years will be crucial in this space, and by 2020, we at least expect to see Waymo open an autonomous ride sharing service, and Google’s Android interface in cars becoming increasingly commonplace.
For more details on Android in cars, click here.