The automotive industry has been known to be slow in innovation. Remember the days you could not start your engine because you forgot to turn off your headlights. How long did it take the industry to come up with the idea of adding one simple alarm system to alert the drivers to turn off the headlights …..10 years? But that trend is changing. The whole industry has wakened up to the fact that they need to come out with new innovations every two to three years instead of ten. What will future car innovations look like? Future cars will be more connected in many ways and they will drive themselves. Automakers and technology companies collaborate on innovation racing to be the first to introduce self-driving cars.
BY JOHN KOON, EDITOR-IN-CHIEF
Most new cars sold, today, are equipped with the Bluetooth-enabled microphones and speakers to allow drivers to talk on the phone hands free. But more features will be added. According to BI Intelligence (www.businessinsider.com) , there will be 380 million connected cars on the road worldwide with built-in hardware enable people to stream music, be alerted of the traffic and the weather conditions. Figure 1. Additionally, auto makers will be able to access data on cars’ performance and allow cars to download software updates without going through a recall process. We start to see these connected car technologies being implemented already. For example, Covisint, a company provides a secure platform to help automakers like Ford, GM, Daimler and Hyundai to connect their ecosystems, enables Hyundai owners to remotely control certain functions of the car via a Smartphone. Hyundai’s recent video promotion features a couple relaxing on the beach in a hot day. When it is time to go home, the owners simply turn on the air conditioner remotely and when they get to the car, it is already nice and cool inside! Connected cars offer drivers additional convenience such as navigation, remote roadside assistance and mobile internet hot spots. But there is a downside. Connections also open up new opportunities for hackers. Experts warn that more attacks will occur and it has been shown how easy for a car to be hacked with someone who is not the owner takes over the control. Security companies such as Symantec Corp. offer solutions to counter hacking. Symantec recently introduced a product called Symantec Anomaly Detection for Automotive which uses machine learning to provide passive in-vehicle security analytics of the commonly used Controller Area Network (CAN) bus traffic. It will learn what a normal pattern should be and take action if abnormality is detected.
A total of 380 million cars shipped will be connected one way or the other. This allows people to stream music, be alerted of traffic and weather conditions. Source: BI Intelligence.
Another future connected car technology being investigated and researched by the National Highway Traffic Safety Administration (NHTSA) is vehicle-to-vehicle (V2V) communication. This is done with partnership of United States Department of Transportation (DOT), the automotive industry and academic institutions. Cars equipped with V2V technology will be able to comununicate with cars nearby to alert drivers of dangerous condition to avoid crashes. For example, a driver will be warned that another car at the intersection is approaching quickly even though it has not yet been seen. V2V makes use of a protocol called Dedicated Short-range communication (DSRC), a two-way, wireless communication with a range of 300 meters depends of the surrounding environment. It operates at the 75MHz band of the 5.9 GHz spectrum allocated by the FCC. V2V technology has the potential of reducing car crashes and costs.
The Race is On to Develop Self-driving Cars
To catch the wave of self-driving car development, technology companies and automakers are rushing to either acquire other companies with autonomous driving capabilities or form new partnership. In July of 2016, Intel, the world’s largest semiconductor company, along with BMW Group and Mobileye N.V. jointly made the announcement to produce “highly and fully autonomous vehicles” by 2021. Additionally, Intel acquired a Russian based machine vision company, Itseez, to support the self-driving effort and Italian based Yotitech to help prevent car chips and systems from failing. In March this year, General Motor purchased Cruise Automation to accelerate the development of self-driving capability. Ford teamed up with MIT and Stanford to focus on the automated driving research. They are also the first company to test self-driving car in a snowing condition. Last year, Ford opened a brand new research center in Silicon Valley to develop innovation relating to connectivity, mobility and autonomous driving. Early this year, Ford tripled its self-driving car fleet making it the largest of all automakers. Google has teamed up with Fiat to further its self-driving efforts. The list goes on.
According to the 2014 forecast done by Lux Research, Inc. (www.luxresearchinc.com) the global driver assist market will grow to $102 billion in 2030. Figure 2. It is interesting to note that very small percentage of the revenue would come from fully-autonomous cars. Executives of various automakers, however, are much more optimistic about setting goals to deliver self-driving by the 2020 time frame. Both Ford and BMW have made public statements of their projection. General Motors, on the other hand, does not want to provide a time frame of when self-driving cars will be available. For consumers, it is very easy to get the wrong impression that self-driving cars will be coming soon because video clips shown in prime time news featuring drivers checking emails, drinking coffee or even watching TV in self-driving cars as if they are watching a TV advertisement. The National Highway Traffic Safety Administration (NHTSA) defines vehicle automation as having five levels. Here is the summary.
The global driver assist market will grow to 2 billion by 2030. This forecast includes all level of autonomous driving. Source: Lux research, Inc.
- No-Automation (Level 0): The driver is in complete control at all times.
- Function-specific Automation (Level 1): Provides one or more specific control functions such as auto-braking.
- Combined Function Automation (Level 2): Provides at least two primary control functions cruise control in
combination with lane centering.
- Limited Self-Driving Automation (Level 3): Vehicles will have full control of all safety-critical functions under certain traffic conditions. The driver is expected to be available for occasional control.
- Full Self-Driving Automation (Level 4): Fully autonomous; vehicles can be driver occupied and unoccupied.
The development of self-driving technology is a progressive process. Without a double, we will see Level 1 and 2 driver-assist technologies to be available in a few years. When will a level 4, fully-autonomous, driverless cars be commercially available is anyone’s guess.
How Does Self-driving technology Work?
The concept of self-driving cars are both fascinating and challenging. Think about what you do when you drive. There is lots of information to be processed while driving. Self-driving cars have to simulate what human do. They rely on machines to process information received and decide what to do. First, machine vision has to know what to look for. Then all the sensors installed on the car have to function well to detect what objects are around the car. Theoretically, a self-driving car uses sensors such as radar, sonic, LIDAR and other means to detect its environment and drive the car according to the traffic rules. Machine learning or deep learning is an important part of self-driving technology development. That is why companies like NVIDIA, IBM, Google and Baidu are so eager to get involved. “Deep learning powered by the GPU is transforming every industry from health care to finance to automotive. In the very near future, an in-vehicle supercomputer designed for artificial intelligence applications will be the safest path forward to enabling autonomous vehicles on our roads, “commented Danny Shapiro, Senior Director of Automotive of NVIDIA, a leader in supercomputing and deep learning. Roborace will host a race of self-driving sports cars in the upcoming months in multiple cities including Long Beach, California, London, Beijing, Berlin, Paris, Mexico City and more. These futuristic, high-performance, electric cars will be driving themselves. Figure 3. NVIDIA and Roborace collaborated on the design of these self-driving race cars which are powered by the NVIDIA DRIVE PX 2. The different teams will compete by developing AI software for the cars. This will be a historical milestone on self-driving technology in general. The race track is a well-defined infrastructure in which deep-learning should perform well.
NVIDIA and Roborace collaborated on the design of these futuristic, high-performance, electric self-driving race cars powered by the NVIDIA DRIVE PX 2. The different teams will compete by developing AI software for the cars.
While there are great benefits, there are still a lot of technological challenges to overcome.
Yes, computer-assist brakes can sense what is in front and stop the car better than you can. The question is when a DUI driver coming from behind at a much greater speed than your vehicle and there is another car in front of you, what would happen? Will your vehicle move to the left, to the right or simply brake as hard as possible. If it decides to move over to the left lane, will the sensors be smart enough to detect if there are cars coming from behind in that lane? No doubt designers will be thinking about how to overcome all the major challenges in front of us.
- How to perfect the self-driving technology?
- What sensor and V2V technologies would work best?
- How to overcome the public fear and prevent another fatal crash relating to “autopilot” in the self-driving mode?
- How safe is safe enough?
The automotive market is wide open with opportunities for everyone and to be the first to come out with commercially available self-driving cars will make history. As self-driving cars need a lot of computing power, software and electronics, technology companies will thrive in the self-driving car growth. Micron, a leader in memory chips, is already projecting a 30%-40% annual growth from the automotive segment. QNX, with its software already running in some 60 million vehicles today, will enjoy additional growth. Expect to see innovative start-ups to change the landscape of self-driving industry. We will continue reporting on the progress of the development of self-driving technologies when we see significant new milestones.