In January 2015, some of the best vehicle autonomy researchers in the world began disappearing from the National Robotics Engineering Center (NREC) at Carnegie Mellon. By the end of the month, fifty NREC staff members had defected from the research institute―fully a third of NREC’s headcount―including many of its top employees and the center’s director.
The erstwhile staff members reappeared just a few blocks away, inside a former chocolate factory now owned by Uber Technologies, which had purchased and renovated the space to house the flagship office of its Advanced Technologies Group. Uber had lured the researchers away from NREC with sky-high compensation packages and the promise of making a tangible impact by helping cars drive themselves in the real world―not just in the lab.
Later that year, Carnegie Mellon and Uber signed a “strategic partnership” to reset their relationship and create a more formal pathway for CMU researchers to engage with Uber, but other players in the automotive space had already taken notice. As audacious as it was, Uber’s bold maneuver was merely the opening salvo in a battle to find and recruit the most valuable talent in the world―autonomous driving experts―by any means necessary.
In recent years, the talent war has escalated. Today, Google’s Waymo division is embroiled in a lawsuit against Uber, accusing former employee and self-driving expert Anthony Levandowski of stealing trade secrets after Uber paid nearly $700 million to acquire his startup, Otto. Traditional automakers have entered the fray, with General Motors paying $580 million to acquire Cruise Automation and Toyota earmarking $1 billion to build a 200-strong research team for self-driving cars. With a $42 billion market at stake, self-driving experts are being courted with the same kind of fervor―and compensation―as rock stars and top athletes.
How Talent Drives Innovation In Vehicular Automation
As discussed in our article on the growing importance of software to the auto industry, self-driving vehicles will fundamentally alter the automotive value chain. Until the recent past, automakers have been hardware powerhouses with a core competency in manufacturing and logistics. But sophisticated software will be equally vital to the cars of tomorrow―for self-driving cars, as necessary as the engine and transmission.
From a talent perspective, the most obvious need of players in the mobility space―transportation networks like Uber and Lyft, automakers like Tesla and GM, and new entrants like Waymo and Apple―is finding experts in machine learning, computer vision and artificial intelligence with the know-how to design the “guiding intelligence” of autonomous vehicles. These are the systems which translate sensor inputs and map data into self-driving capabilities that demonstrate enormous reliability across a wide variety of conditions.
But the need for talent doesn’t end there. Self-driving cars are at the nexus of four interrelated technological trends, each representing a horizon that must be pushed to the limit to create a superior mobility experience that will win consumers in the next decade. In the rest of this article, we review those technologies―connectivity, autonomy, shared mobility, and electrification―to help automotive stakeholders prioritize their search for talent.
As is the case with all applications of machine learning technology, self-driving capabilities are a product of their data. With higher-quality data, autonomous vehicles can make better, more reliable decisions about where and how to drive.
Most of the mission critical data required for cars to drive themselves will arrive from a robust complement of onboard sensors―for example, fully autonomous Teslas have eight cameras, twelve ultrasonic sensors, and forward-facing radar. But to maximize performance and safety, autonomous vehicles will need to connect to additional data sources.
At a base level, this involves connecting to GPS guidance systems and cloud-based traffic applications like Waze to help vehicles generate optimal routes. Automakers are already on top of this trend, and Gartner estimates that nearly 250 million vehicles will be connected to the Internet by 2020. But reducing latency and reaching the highest standards of safety and comfort implies that self-driving cars will also share data and communicate with other vehicles, and with the infrastructure and roads around them. For example, smart traffic signals could interact directly with vehicles―the city of Atlanta even plans to build separate roads for self-driving cars, complete with sensor-embedded road signs and parking meters.
Self-driving cars will consume data from a wide variety of sources, implying that expertise in connectivity and the Internet of Things will be essential in this field. It will be just as important to retain top security experts who can detect vulnerabilities and limit the risk that connected vehicles are hacked or exploited.
Having ingested a wide variety of data from internal sensors and external sources, self-driving cars must transform it into route and control guidance. The algorithms that produce autonomy are the most fundamental component of driverless vehicles, and autonomy is the capability at the heart of the talent war.
The online education platform Udacity recently launched a “Self-Driving Car Engineer” nanodegree taught by Sebastian Thrun, a professor at Stanford and one of the godfathers of self-driving car research, and its curriculum provides a glimpse into the many software competencies required to produce vehicular autonomy of any degree. Essential areas of knowledge include deep learning, computer vision, sensor processing and fusion, localization, and control―each with in-depth submodules. Taken together, these skills enable engineers to design systems that can recognize traffic signals and signs, keep lanes, adjust to adverse weather, react to the flow of traffic, and preempt potential collisions.
It is notable that a number of Udacity’s course modules are sponsored by companies researching self-driving car technology, including Mercedes-Benz and Uber, implying that major mobility players are so eager to lock down autonomy talent that they are willing to scoop up engineers straight out of school.
3. Shared Mobility
Research into autonomous vehicles has been accelerated tremendously by the growth of transportation networks like Uber and Lyft, as well as car-sharing services like Zipcar and Car2go. In particular, Uber and Lyft are banking on the fact that autonomous cars will substantially reduce the variable cost of their services by eliminating drivers, making ride-sharing competitive in cost and convenience with owning a car.
The popularity of Uber and Lyft, combined with cultural factors that have reduced the value of car ownership for younger demographics, have made mobility players of all kinds sit up and take notice. Tesla is one automaker that has embraced the idea that the future of automotive mobility might involve far less ownership of vehicles, announcing plans for a “Tesla Network” of self-driving Teslas that could pick up other passengers during their down time. GM has also taken an aggressive approach, acquiring the assets of ride-sharing startup Sidecar and partnering with Lyft to fund their expansion and pursue a number of strategic initiatives, while Ford has announced its intention to launch a fully autonomous vehicle for ride-sharing by 2021.
Whether they want to supply vehicles for ride-sharing networks or create their own networks, mobility players will need to invest in talent to understand challenges in the space, from smart routing while carpooling to maintenance and passenger safety.
Self-driving cars with internal combustion engines will exist well into the future, especially for long-distance applications. But as battery technology improves in range and cost, the next generation of autonomous vehicles is more likely than not to be electric.
The reasons for this tie back to the other technological trends that have catalyzed autonomous vehicle development. For one, electric vehicles have the potential to be easier to maintain, because they consist of just three main components―battery, inverter, and electric motor. At maturity, they could also be easier to refuel through wireless technology like induction charging, making them well suited to the intense usage patterns of ride-sharing.
Electric cars are also easier for computers to control, and can furnish reliable power to the array of sensors that autonomous vehicle use to gather data and control their motion. Already, 58% of light-duty autonomous vehicles are built over an electric powertrain, while an additional 21% use a hybrid powertrain. With this said, until battery technology enables electric vehicles to match the range of gasoline cars, hybrid engines will bridge the gap, offering both range and superior compatibility with the unique needs of autonomous vehicles.
Good Self-Driving Talent Is Hard To Find
The talent wars in vehicular automation have just begun. The demand for technical and business talent in the four areas outlined above will grow as organizations across the entire mobility space increase their conviction in the rapid progression of the technology and legal norms that currently gate self-driving cars from widespread adoption.
Securing talent with deep expertise in autonomy is essential to producing autonomous vehicles, but mobility players must not stop there. The winners of the talent wars will need to look across the entire stack of competencies required to produce safe and high-quality autonomous driving experiences, employing creative tactics to lock down talent and secure their road to sustainable growth in the next era of mobility.