No one can have failed to notice that the rate of technological change in the automotive world is accelerating fast. While technology has for some years now increased car safety in the form of ABS, traction control and airbags, it’s only relatively recently that more advanced technology has started to be introduced and, in some cases, become standard.
These systems are known as Advanced Driver Safety Systems or ADAS and encapsulate everything from features such as adaptive cruise control, all the way up to full autonomous driving. There are also ever more sophisticated advances in in-car entertainment.
Imagination’s IP, from its PowerVR graphics and video designs to its MIPS CPUs is already widely used in automotive, and this looks set to increase further over the next year and beyond. Drawing on this extensive knowledge we present our top 10 insights for the year ahead in automotive split into four sections; security, autonomy, infotainment and technical.
OK, let’s drive in:
- The first real car hijack will happen in the real world
As cars get ever more reliant on technology and connectivity, the threat of wireless hacks gets ever more real and the chance of the first real car hijack with occupants inside is a distinct possibility in 2017. The feasibility of these attacks has been demonstrated numerous times. A Jeep was hacked in July 2015, driving it off the road, while only recently a team of Chinese hackers recently accessed a Tesla Model S P85 and 75D, applying the brakes remotely. Considering the hacker ‘kudos points’ it could earn it’s too much of a temptation for it not to occur. It’s a scary scenario and it will be beholden on car manufacturers to look to solutions that offer security at their core in order to reassure customers and ensure their safety.
- Hardware-backed security for cars from ground up
The car industry knows it has a security problem. With no standardised security platform today’s security is purely software-based. However, unless it’s backed with security in hardware it can be cracked. This has already happened to the ECU infrastructure of a leading tier 1 manufacturer. In 2017 car manufacturers will develop their ECUs with a hardware root of trust plus secure boot and crypto libraries making it a whole lot more difficult for hackers to have a field day.
- Self-driving car trials will begin in earnest
In 2017 we will see trials from major car manufacturers on both Level 3 and even Level 4 autonomy, with an aim to deployment by 2021. (Go to this site if you want more info on the differences between the Levels) Level 3 means the car can drive itself under certain conditions and only requires driver intervention if necessary. Volvo’s ‘Drive Me London’ plan announced plans for 2017 trials, while Ford will begin testing in the UK and Germany too. Uber is pushing for offering self-driving cars (with a driver inside for safety) and while these have been removed from the road by the California Department of Motor Vehicles, they are sure to be back at some point in 2017. Trials in China have already started and will continue next year, while in Japan a trial of self-driving cars used for deliveries will begin in March 2017.
- Neural network hardware acceleration will become a requirement
For cars to reach higher levels of autonomy, the leading computer-vision processing companies are looking to make use of deep learning to make autonomous cars make decisions faster and more reliably. Convolutional neural networks (CNNs), optimised for finding detail in specific areas of an images and variants of deep neural networks, will power these deep learning systems. Fast, efficient hardware is required to run these CNNs and the highly parallel compute abilities of PowerVR GPUs make them a good fit. In 2017, GPUs will offer the performance required at the required power levels for hardware acceleration of CNNs.
- The average new car will have four cameras built-in
Aside from autonomous driving automotive safety will be greatly increased in 2017 as most new cars will come equipped with four new cameras. Rear view backup cameras will be mandatory in the US by 2018, but next year cars will also offer front and side cameras to provide surround view for the car. While these will be used for ADAS features such as adaptive cruise control and automated emergency braking, they will also help drivers avoid damage to cars through contact with kerbs.
- Cost of LIDAR solutions will drop significantly
Light Detection and Ranging (LIDAR) uses pulsed lasers to build up a highly accurate 3D picture of the world and is touted, along with cameras, as key to enabling self-driving cars. However, it has traditionally been a very expensive option costing as much as the rest of the car again, making them impractical for mass produced cars. This will begin to change in 2017 with companies announcing a chipset costing less than $50 for volume orders. This won’t equate to the costs of a full unit being added to a car, but it bodes well for the affordability of units that will power cars to meet the self-imposed deadline of 2021 for full self-driving cars from manufacturers such as Ford.
- Apple and Google to increase presence in infotainment
2017 will see Apple Google make big inroads into their battle for the eyes of consumers with continued development and availably of Apple Car Play and Android Auto. Thanks to these, consumers can bring much of their smartphone experience into the car, safely and conveniently, offering up key smartphone functions on a touchscreen optimised for in-car use. While Apple’s self-driving car project is rumoured to have gone off-track CarPlay is looking good for 2017 while Android Auto, which offers similar functionality and a raft of Google services, will also enjoy wide support.
- AR and VR technologies rolled out in buyer experience and maintenance
Thanks to the increased use of AR and VR, those looking to spec up a brand new car in 2017 will be able to do so from the comfort of their sofa. Just strap on a VR headset and they’ll be able to see how their exterior will look like in different colours and compare the differences between the different fit and finishes available. Once they buy the car they’ll be no need to reach for the manual – an AR app will bring it to life. AR will show them how to operate the controls and the infotainment system, while engineers will also be able to use AR to talk them through repairs on a car, making maintenance and training easier and more engaging.
- ECUs will move from 8/16-bit to 32-bit
Currently most cars make inefficient use of computing power. The average car has 50 or so microcontrollers and microprocessors but many of these are typically used for less than one per cent of the time. This means that there are a multiplicity ECUs required for different tasks, from basic car operations such as lowering the windows to managing the keyless access. This is a highly inefficient way of using the available processing power. In 2017 car manufacturers will start to move to 32-bit microcontrollers with hardware virtualization enabling new levels of performance and security. Thanks to hardware virtualization a single CPU will be able to handle multiple tasks safely and securely, driving both costs down and even reducing weight, helping to increase efficiency.
- Introduction of automotive Ethernet will start movement towards a centralised car computing platform
While the car’s ECU/CAN bus architecture has served the industry well for the last 15-20 years, given the requirements of autonomous driving this architecture is insufficient for the to deliver the safety features and functions required in the future. In 2017, the industry will start to move towards an open platform, through the use of APIs enabling the software world to make full use of the car. While this will take a number of things to come together, it will start with the introduction of automotive Ethernet, delivering a standardised high-bandwidth internal data transfer platform and turning the car from a hardware platform with some software, to a software platform attached to some hardware.