Despite a strong start, the National Highway Traffic Safety Administration (NHTSA) has declared that the permissions which prevail in Nevada, Florida, and California for automated vehicles will only be applied for testing purposes and none other, until proven completely safe. For the same reason, European legislation – and the various disparate versions in each member state – has thus far not taken a call on what sort of automated vehicles can be permitted on public roads for personal use.
Although the purpose of the automated driving initiative is aligned with taking road safety to the next level, the biggest nightmare for all stakeholders involved – OEMs, suppliers and legislators alike – is the accident liability in the event of an automated vehicle colliding with a pedestrian or another vehicle. This could be caused by various factors: glitches in various ECUs, false-negatives (missed warning), hacking and related cyber-security threats, wrong override by driver, and failure to recognise a corrective override command from a driver are just a few such examples. As a matter of fact, these convolutions have already delayed OEMs from launching automated vehicles. Some, including Mercedes-Benz and Audi, have developed production-ready driverless cars, but have chosen to disable the functionality due to the legislative and real-world imperatives.
Consumer appetite for vehicle automation can be classified in terms of what extent of vehicle automation they may digest. When offered freebies worth a few thousand dollars, some consumers opt for infotainment features, some for safety, others for extended warranty, while another set of consumers prefer a mix of all of these. In the most general case, it is only logical to state that a large fraction of people will still like to drive their cars, letting it drive itself only under certain conditions – and one can apportion anywhere between 10-90% of driving time to the term ’certain conditions’.
The Mercedes-Benz E- and S-Class vehicles of 2014 have proven that they can drive themselves in adverse conditions without the presence of a physical driver inside. BMW’s automated vehicle will also ensure that you are pulled over to a hard-shoulder if you faint or fall asleep while behind the wheel. Volkswagen’s Temporary Auto Pilot (TAP) expects the driver to pay complete attention to the road, even if not actively driving the vehicle, expecting the driver to take over at any point, when prompted. Audi’s piloted driving feature will let you get out of the vehicle, right in front of the lobby at a hotel, while you can use a smart-phone app to park and retrieve it. The same feature is extended to a piloted driving mode, when the driver hands over the driving tasks to a virtual co-pilot, i.e. the vehicle itself. Most of these innovations can be done if state-of-the-art ADAS and connectivity features are leveraged to leave no stone unturned. At the risk of sounding too generic, real-world challenges require smarter in-vehicle systems – both for ADAS and connectivity – to make the vehicle see, feel, and decide like a real driver would. A classic example of this would be speed-bumps and potholes: while speed-bumps and ramps can be recognised with a smart forward-looking camera aided by radar or ultrasonic sensors, potholes, an open manhole or any other dug-up area in the path of the vehicle may not occur to the vehicular sensing systems so very readily.
Google it
Much has been said and written about the Google Self-Driving vehicle project. Impressive, as it is, for a non-automotive participant to gain sufficient attention for a project outside of their core competence, one would wonder whether consumers would trust the likes of a Google in any manoeuvre other than navigation, app-based POI information etc. Although traditional OEMs such as BMW, Audi, Mercedes-Benz, General Motors, and others are yet to make a firm statement on the confirmed launch of such vehicles at a decided price, the industry is left wondering what exactly Google is doing here. Traditional OEMs are intent on semi- or highly-automated modes while Google is going all out to create vehicles that do not need a driver inside.
For Google, it is clearly about connecting the dots. They have connected navigation and location-based services either through Google Now and Google glasses. They have the ability to liaise with channel partners to put the internet into cars and ensure that occupants spend more time using some sort of Google application. All it takes is a vehicle that can drive itself so that people maximise their time using Google’s other products. Take the example of Google Glasses: wearing them while driving is a punishable offence in the UK. So Google is trying to shift the driving responsibility from the driver to an automated driving module, so that the human driver is downgraded to an occupant of the vehicle, free to use Google Glasses, at least in the eyes of the current law. Legislators and OEMs may be trying to ensure that self-driving vehicles reduce the probability of road crashes, but Google, on the other hand, is looking at the same as a potential business opportunity.
The best business case for a Google vehicle is to partner with an OEM that does not have an action plan to launch an automated driving vehicle in the near future, or with a manufacturer that does not boast the best ADAS and connectivity systems. Google’s solution should be a packaged bundle that takes a two-pronged approach in both active safety and in-vehicle connectivity. One possibility is to fit an optional Google-X add-on to a capable vehicle model, on the condition that the OEM would avail of an infotainment package from Google. The much ado about software-over-the-air and firmware-over-the-air need not result in nothing, if this proposition is a reality. Thus, the idea would be one of Google making Android-like automated driving modules, compatible with a specific list of cars.
Should all pieces fall into place, the future will see a set of OEMs who offer proprietary automated cars and the likes of Google fitting certain other cars with an automation module and Google-powered infotainment unit, lest the occupants be bored to death. No one will be surprised to see Apple entering this space, considering the popularity of the iPhone in markets where automated cars are likely to be launched first. Indeed, the computer giant has already established an automated vehicle research arm.
Google and Apple are two of the biggest names to have revolutionised digital media and now both are likely to provide an ‘infotaining’ environment on the move. In the automated driving space, much like their electronic advent, one can expect Google to be hardware-agnostic, creating adaptable platforms – perhaps even open source – whereas Apple may be looking to leverage its technical prowess in sleek hardware development and purpose-built software.
Capable driving, responsible driver
‘Can I order a vehicle online and expect it to drive itself to my garage?’ has been the question in the minds of many consumers. Such a scenario may not be possible in the foreseeable future, but what one can expect in the next five years is a vehicle which is capable of driving itself but still holds the driver responsible for the driving task. Fully-autonomous cars do not require a human to be present inside to drive the vehicle. Such a vehicle will then not need the same architecture as a vehicle of the present day: steering wheel, brake-pedal, throttle, gear-shift etc., can all be removed and the vehicle would have more space for comfort, convenience, and infotainment features throughout. And if such a vehicle would not need a driver inside, should there be one outside? This is likely to pave way for a whole new business model for driving-as-a-service.
Similar to software-as-a- service, which works on a thin-client mode, hardware at a remote site needs to be compatible only with the interface rather than capable of running the background codes – thus vehicles of the future may only need to provide inputs to a driver seated outside of the vehicle.
But what is the business case for such a model? Governments are still not sure if they should allow autonomous vehicles for public use. Consumers themselves are divided on whether they would feel comfortable making their child sit in a vehicle that drives itself. While insurance companies should ideally support the idea of automated driving – as such vehicles are lesser likely to be involved in a crash – there is no single definition of automated vehicle architecture or a set of type-approved vehicles that are automated. The overall sentiment is that human drivers are more trusted by one and all, rather than machine-vision driving a vehicle. In such a case, as mentioned above, the driver need not even be present inside the vehicle to drive it.
Stakeholder cooperation needed for such a scenario would include road-operators, infrastructure providers (smart-grid, V2X, among others), telecom operators, telematics service providers, insurance companies, and fleet operators. When these entities come together and help the legislation formulate the most practical and most appropriate set of regulations, we will then have cars that drive themselves, under someone’s monitoring. Even in tunnels, such vehicles can follow non-automated manually driven vehicles, if concerns exist about connectivity losses. Whatever may be the outcome, there needs to be multiple redundancies such as duplicate telematics boxes, dual connectivity modules and V2X on top of telematics and ADAS, to name a few.
Prana Tharthiharan Natarajan, is Team Leader – Chassis, Safety & Driver Assistance Systems, Automotive & Transportation, at Frost & Sullivan. Frost & Sullivan works with MIRA to study the market and technology for autonomous driving.
