Analysis: CBR speaks to Nissan, Nvidia and other industry heavy weights on bespoke connected car technologies changing the motor sector.
Smart, connected, autonomous or driverless, the transportation industry is being driven through one of its biggest technological shifts.
Despite drivers’ concerns, connected car technology is arriving in different shapes and forms, from sensors, cameras or nodes, to full smartphone integration, head displays, high-digitalised infotainment systems, 3D mapping systems, and others.
Industry giants have recognised that tomorrow’s business will be based around connectivity and intelligent services both in and out of the car.
This has led, for example, Ford’s CEO Mark Fields, to announce last month that the 113-year-old company is no longer just an auto manufacturer but an auto and mobility organisation.
He said: "Transportation is at the cusp of a revolution, and it is inspiring evolution at Ford."
With the annual Geneva Motor Show undergoing in the Swiss city, makers from all over the world drove in their latest car technology and their vision of how cars will look and serve humans in the future.
Dr. Kevin Curran, senior member of the IEEE said that cars will become more integrated with national intelligent transport infrastructures and systems, however, a crucial barrier to success is still the lack of trust and collaboration between the major auto manufacturers.
He told CBR that glimpses of future intelligent infrastructures can be seen at the moment in initiatives such as one in the UK where in Birmingham City, IBM are helping to analyse big data to help understand parking patterns in order to better manage congestion.
"They deployed ultra-low-power wireless sensors in roads and offered an accompanying app for drivers to get real-time availability and prices for parking. Common sources to manage traffic include road sensors, video cameras and GPS updates from public transport. Smart cars will be expected to integrate with such infrastructure in the future."
Dr Curran said that crucial components of the future will be the mobile networks, ad hoc (car to car -C2C) networks, vehicles to/from road sensors and satellite communications. "We can expect a significant portion of the internet to be consumed by vehicle communications."
According to Dr Curran, C2C technology will be one of the key enablers of the whole car industry in the future.
Applications of C2C include for example, on-board C2C technology automatically sending vital details to the emergency services in case of a car crash. This data could include time of collision, GPS location, vehicle description, vehicle licence number and registered owner. "This might save crucial moments in life-threatening situations."
John Smith, principal solution architect at Veracode, told CBR: "It takes skilled people, processes and tools to streamline the creation of secure software, all of which need investment and take time to get right.
"Even the most mature software development organisations will struggle to create perfectly secure software, so it is essential that security is an integral part of the ongoing lifecycle of connected cars.
"It is important that the manufacturers focus on the long-term implications of software security for their customers and not just on the short-term gains that a new or improved feature can bring them."
Ultimately, autonomous and driverless technology is making its way into the consumer space and at the Geneva Motor Show this has become evident with companies like Nissan, Audi and Ford strengthening their takes in this space.
Nvidia’s Drive PX AI technology
AI deep learning making cars extra smart
Attending the show was Nvidia‘s director of automotive business unit, Danny Shapiro, who described the firm today as being a computer company developing automotive solutions.
"The automotive part of our company is relatively small but it is the fastest going part of the company," he told CBR.
"The evolution is from purpose build or hard-coded kind of algorithms, to a more open system where we are using Deep Learning and AI to solve the problem and letting the computer system learn and get smarter over time and do software updates, like people do in their smartphone, which is going to extend the car lifetime and add functionality.
"The car is going to get better as they own it, versus of just being a fixed point in time that never gets better."
Targeting this, Nvidia has designed two major smart car technologies: Drive CX and Drive PX.
Shapiro said: "Drive CX is a cockpit computer, it is the graphics focus aspect, the infotainment screens, the instrumental cluster, head of display, etc, wherever you have pixels in the car, Nvidia is running software on our graphics processors.
"We have over ten million cars on the road today that have Nvidia processors driving those different applications. We are going to see a lot more cars on that infotainment and digital concept."
Secondly, the Drive PX is a self-driving car supercomputer, and comes as a hardware and piece of software. It contains a powerful processor that scales from one to four processors and can deliver up to 24 trillion operations per second.
"It is a true supercomputer that has the performance of 160 MacBook Pros. It is designed to process data that is coming in from sensors, could be cameras, radar, ultrasonic, any combination, and essentially we are trying to interpret the data coming in and understand exactly what is going on in a 360 degrees around the car. That is really the brain of the self-driving car," he said.
The system is capable of recognising other cars, trucks, cyclists, pedestrians, signs, and other moving or standing objects on and around the road.
In the autonomous and driverless car space, artificial intelligence (AI) also plays a staggering role. Shapiro said that in the AI, Deep Learning is a key aspect to understanding the AI concept.
Deep learning is used in the research community and in industry to help solve big data problems such as computer vision, speech recognition, and natural language processing.
NVIDIA provides tools and libraries to power GPU-accelerated machine learning applications in the cloud, data centres, workstations, and embedded platforms with the Deep Learning SDK.
Shapiro said: "The system needs to be trained, basically build a model of how the human brain functions. Once we defined that neural network, we have to teach it. That will happen off-line, in the cloud, and will feed it millions and millions of images.
"In terms of vocabulary, [the system is trying to define] what is a car, a truck a bike, what is pedestrian, a tree, a sign, etc, and that could take weeks or months of data processing running on GPU. A graphic processor runs it ten to 50 times faster on a GPU."
He said that what we are now seeing, is that in the GPU, what used to take a month to train could take a day and that gives developers, automakers and others the ability to refine their neural networks, and train them with more data.
Once the network has been trained, then it is deployed in the car. "We download that neural network model on to the Drive PX in the vehicle and that runs in real time. As the camera sees, or other sensor data comes in, we can determine in a fraction of a second what the cameras are seeing or the sensors are detecting.
"We can then understand everything around the vehicle. Based on that, you figure out where we are and where we want to go, what is the next step forward, accelerating, breaking steering left, right, etc."
Shapiro said that all that process needs to take place on board and there is no way analytics can go offline, "to the cloud, edge, whatever, to be able to make those 50 critical decisions in a fraction of a second reliably".
"That is why such a powerful supercomputer is basically required on board, to be able to process and make those decisions in just a fraction of a second," he said.
Nvidia is currently working with over 70 partners in the car segment, from automakers like Audi or Tesla to research organisations like the MIT.
Bringing connected cars to the masses
Elsewhere, Nissan is also looking into the field of connected and autonomous cars. Nissan Europe‘s Richard Candler, GM for advanced and CCL product strategy, explained how the company is planning to bring the autonomous driving experience to the masses in the Qashqai 2017.
"This car can drive down a lane and steer automatically. It can also manage the distance to other cars by accelerating, breaking," he told CBR.
Candler said that in 2017 the company will roll out the version 1.0 of its autonomous driving technology, with the 2.0 version expected for launch in 2018.
"This is when the car will be able to change lanes, understand intersections, etc. And then 3.0 [will come] in 2020, this is a car that can basically drive itself in almost every situation.
"With those three systems, the driver is still in control, but they can delegate the task when they do not want to be driving. This really opens up an enormous opportunity for the user to decide when they want to drive (…) whilst taking away some of the pain points of driving."
Candler said Nissan is currently not looking into the driverless car spectrum as the demand is not out there yet. "Past-2020, things might change, people might get more used to the car taking more control," he said.
"What we found from our research with customers, is that actually people do not want to hand over the control of the car completely. What they want to do is to define certain situations where they want to have that."
Beyond in-car connectivity
Nonetheless, connected cars are not just about the data exchange and analytics that happen within the car itself. Vehicles are becoming part of a wider connectivity mesh that will turn roads smarter and more efficient.
For example, IoT cloud provider Kii, is currently developing a smart traffic system in Italy, with other locations being looked at. "In the UK we are still talking to people," Kii’s Anthony Fulgoni, VP for business development in EMEA told CBR.
Explaining how the system works, he said: "The first thing [needed] is to have a device sensor in the road that is cost effective. You do not want it to be expensive.
"[To deploy the system] effectively you drill a hole on the ground, pop this thing on the ground – the battery life is about six years – and you leave it. On the road side, there is a small device that can be in a cabinet or lamp post which does the RF connection between the road sensor and then the internet. That then links into the cloud.
"Once you have several sensors on the ground, you cover all lanes, collect that data, pass it back to the internet, and then start doing big data analytics on it."
Recently, Nissan has also done research into the car working within the city [pictured above], in partnership with architecture group Foster + Partners, and came up with the concept of turning the car into the "petrol station of tomorrow".
Candler said: "The point of the study is really to look outside the vehicle and think about how buildings will evolve. We believe autonomous cars and electric vehicles (EVs) are an answer for problems such as pollution."
The study shows a concept idea where a battery would be embedded in the car and charged wirelessly while the vehicle is parked.
"The idea behind autonomous wireless charging this is that you can have, for example, two wireless charging points in a street and then the cars could autonomously move around during the night, after recharging.
"The infrastructure is much more simple for charging in a city environment. In the new developments it is quite simple [to deploy such technology]; on existing roads it is a bit more complex.
"The architects view was that if you could add one or two charging points and share them between the whole street, this becomes very viable."
Candler said that today’s cities are not a very friendly place for the car, and big changes are needed.
He said that EVs will solve the air pollution problems in our cities, and autonomous driving will be a big part of that as well in terms of managing congestion, and also being able to use space more evenly, by for example, sending the car to park outside the city, or parking in tighter spaces, and call the it back when needed.
What does the future hold?
In the UK, connected cars are not only expected to add £51 billion to the economy (up from today’s £1 billion), but could also cut serious road traffic accidents by more than 25,000 a year by 2030, according to KPMG.
The company predicts the production of connected cars in the UK to rise from just under a million in 2016 to nearly two and a half by 2030, with 100% connected car market penetration to happen in 2026 (2016 stands at 55%).
Despite drivers’ concerns around the safety of the connected car, KPMG predicts that between 2014 and 2030, the technologies in this space will save over 2,500 lives in the UK, and prevent more than 25,000 serious accidents in the space timeframe.
Yet, Dr Curran said that a risk associated with rolling out technology in smart cars as opposed to other platforms is the potential of distraction leading to accidents due to poor design or malfunctions.
"Technology experts outside of aviation and medical products tend not to follow stringent testing methodologies but lazily rely on fixing problems as they arise. Therefore a misconfigured service in a fast moving smart car can lead to death.
"The motivation to build rigorous and secure systems should be there because it is quite possible that all involved in its design could be held liable if a defect caused or even contributed to a collision."
Nvidia’s Shapiro said: "In terms of mass adoption [of driverless cars], the key to get it faster, is going to be around regulatory. Different countries will progress at different rates.
"What we will see, is the progression of certain environments, that might be a private property, college or corporate campuses, islands, etc, where full autonomy is allowed. Once the technologies are proven to the regulators and the consumers then we will see more mass adoption."
Finally, it is electrifying to think that, despite all the problems encountered and technology developments that still need doing, connected cars, from semi to fully autonomous, will change the way humans move around and re-shape cities around the world. One thing is for sure: nothing will ever be like it is today.