Driving into the future

Automatic cars are the next target for IT firms

Chip maker Intel announced last week that it is taking Israeli firm, Mobileye, by paying about $15.3 billion in an all-cash deal. This may seem a high price since Mobileye’s 2016 revenues amounted to just $350 million. But Mobileye's revenues grew by 49 per cent last year and it holds more than a 70 per cent market share in visual-processing software for self-driving vehicles. Intel makes “artificial eye” chips for drones and cars. Mobileye supplies sensors and software to 26 large automobile manufacturers and its proprietary algorithms compress data from sensors and camera into small, swiftly-analysed packets. All images are fed into Mobileye's neural network to create massive data sets. Machine-learning is used to annotate the images to “teach” vehicles to identify road markings, signage, other vehicles, pedestrians, etc. Mobileye received poor publicity last year when a Tesla car suffered a fatal accident yet it is the standard-setter at visual-processing. Intel-Mobileye is also entering an alliance with BMW, the German carmaker, to process locational data and create real-time updates on traffic density, road hazards, weather conditions, etc. 

Intel is not alone. Every semiconductor major is now heavily invested in automated transport. Qualcomm paid $38 billion last year to acquire NXP, a specialised chip maker that makes advanced driver assistance systems (ADAS) and contributes to car networking and vehicle cyber security. NVidia is in a key collaboration with Tesla and a big supplier to other car manufacturers. After smartphones, this is the next major growth area for semiconductor companies. Specialised chips are needed in data centres to train automobile neural networks and chips are needed in vehicles to execute data processing and control cars on the move. Apart from these “hardware guys”, Google and other networking giants are also heavily invested in autonomous vehicle technologies.

As a consequence, in the future, safety standards will rise: Driverless vehicles have quicker “reflexes”, a 360-degree vision and, potentially, the ability to stay in constant communication with other driverless vehicles. But major legislative changes are required to reset insurance liabilities and review safety standards before driverless cars can become common in a crowded urban environment. By 2020, it is expected that major automobile markets like Japan, the US and the EU will have worked out the new legal frameworks required for autonomous vehicles. America's National Highway Traffic Safety Administration has a six-level standard for autonomy where level zero is completely human-controlled and level-5 is totally driverless. A Level-4 vehicle can operate without supervision in almost all scenarios but a driver with specific training must be capable of taking over. Most vehicles are now at Level-2 and Level-3, while the pioneer, Tesla, is at Level-4. Many auto-majors aim to have Level-4 vehicles ready to roll out by 2021.

The disruptive potential of driverless vehicles is immense, especially in the commercial segments. Driverless trucks operating outside urban limits — such as tractors and combine harvesters, construction & mining equipment — could go driverless quicker since they operate in less crowded environments. As driverless vehicles become more common, the entire transport ecosystem will see a radical overhaul. Driving would become a largely redundant skill, parking lots could be automated and clustered differently and Uber-type models of ridership could become normal. Vehicles could operate 24x7 with pit stops for maintenance. Commuters could work out of mobile offices with car interior designs entirely changing as the steering wheel disappears. 

The IT industry has entered an autonomous vehicle-design phase with few preconceptions and no emotional or financial commitment to the design legacy. Conventional automobile manufacturers could well find themselves rendered obsolete unless they adapt fast to this new reality. At the same time, automobile manufacturers are used to managing very complex, long value chains and, thus, good at building relationships with multiple suppliers and vendors. Alliances like the Intel-Mobileye-BMW one could well become the new normal as driverless vehicles evolve. It will be fascinating to watch two very different industries match and meld their management styles to develop these new technologies.