Cruise balances safety and experimentation in pursuit of the perfect self-driving car
At first glance, the concept of a self-driving car seems deceptively simple. Yet after a decade of bold claims and huge investments from Silicon Valley's most influential organizations, few people understand the difference between the idea and its implementation.
Mohamed Elshenawy, Senior Vice President of Engineering at autonomous vehicle maker Cruise, spends his working life grappling with the difficulties of building a self-driving car: "One of the most common questions friends and family ask me is: 'Why does it take so long to solve the problem of autonomous driving?'"
“We are not training the human brain, which has evolved over hundreds of thousands of years to subtly perceive the world and predict actions in its surroundings. In essence, we are building an AV brain that can do all of this without stagnating or making silly mistakes, without getting tired, bored or distracted. To do this, we are compressing hundreds of thousands of years of human intelligence into just a few years”
Founded in 2013, Cruise was acquired by GM in 2016 for $1 billion. Today, more than 2,000 people work at GM's independent subsidiary, including an engineering team of about 900. Cruise has completed more than 3 million km of autonomous testing.
While some autonomous vehicle contenders chose to test their technology on quieter roads, hoping to solve the "easier" driving conditions before tackling complex urban driving, Cruise jumped straight to the deep end, into the hilly streets of its home city.
In theory, once you solve urban driving, with its construction zones, pedestrians, cyclists and chaotic traffic, it becomes easier to scale and expand the service elsewhere. Cruise has organized its engineering teams around the concept of psychological safety.
“We must be prepared to accept risks within the framework of our safety, taking into account different points of view and new ways of doing things. This experimental mindset is part of a new paradigm shift we are seeing in companies that use machine learning, which are not merely optimized by machine learning but defined by it”
According to Elshenawy, engineers are trying to do something that has never been done before, which makes it an applied experiment in many respects. Psychological safety allows us to rethink how we view failure and leaves room for innovation and personal growth. Accepting a calculated risk is not a failure; it is an opportunity to learn.
The reality is that engineers must have more "failed" experiments than successful ones in order to extract valuable lessons that will help them achieve the next win.
In June 2021, Cruise announced the start of pre-production of its Origin autonomous electric shuttles. The company is developing the concept of futuristic autonomous passenger shuttles with no steering wheel, pedals or driver's seat. In early 2020, the first shuttle was unveiled, featuring a symmetrical body, large sliding side doors and two long seats facing each other.
The company recently received permission to carry passengers without a safety driver on board. Full-scale serial production of the shuttles is planned to begin in early 2023.
“We must balance cutting-edge innovation without compromising safety. These parameters and safety requirements are easy to view as limiting factors, but I have found the opposite. They have pushed us toward a more creative approach and a rethinking of the traditional development process. We have also invested heavily in integrated tools that come together to form an artificial intelligence and robotics platform”
With full-scale serial production starting, the company plans to build "tens of thousands" of vehicles, but it is not yet known over what period. In recent years, Cruise has attracted frequent investment: $2.25 billion from SoftBank in 2018, $1.15 billion from GM, SoftBank and Honda in 2019, and $2 billion from Microsoft in early 2021.
At the same time, some experts believe the company is well behind its main rival Waymo, the Google parent company's self-driving technology unit, which at one point also developed a concept autonomous car with no steering wheel or pedals (in 2015, such Google cars first hit the roads, but the project was shelved in 2017).
Although Waymo already launched a commercial robotaxi service in Phoenix, Arizona, without safety drivers in October 2020, the company has raised investment only once, at $3 billion.
