Waymo, Alphabet’s autonomous vehicle division, is accelerating its expansion across US cities with ambitions to revolutionize transportation. As driverless cars proliferate, though, complications do as well. From public unrest to regulatory obstacles, there continues to be a lack of legal framework surrounding the operation of autonomous vehicles. The future looks uncertain as several Waymo vehicles were set ablaze in riots in Los Angeles against increased deportations.
History
Founded in 2009 by Sebastian Thrun and Anthony Levandowksi, the Google Self-Driving Car Project drew on research from Stanford and leveraged Google’s unmatched mapping and data infrastructure such as Street View and internal navigation tools to construct its early autonomous vehicle systems. Dubbed Project Chauffeur, Google’s autonomous car division made strides from 2010 to 2015, obtaining the first autonomous license, securing a patent, unveiling prototype, and conducting its first public, fully driverless ride in Austin, Texas.
Project Chauffeur was rebranded in December 2016 to Waymo, a subsidiary of Alphabet Inc. Waymo, which is derived from “a new way forward in mobility”, then partnered with Fiat Chrysler to test self-driving technology on Pacifica Hybrid minivans. After copious testing and a legal battle with Uber, Waymo expanded its fleet with Jaguars and more Pacifica minivans while it continued to vertically consolidate, reducing costs by up to 90 percent by increasing in-house production. In 2020, Waymo launched Via, which expanded its interests to freight and logistics. In the following years, Waymo experienced a change in leadership, announced and executed on a roll-out in cities like San Francisco, Phoenix, Austin, and Los Angeles, partnered with Uber, and exponentially increased funding to over $11 billion.
Technology
Waymo utilizes a variety of complex technological components to ensure safety and accuracy. Google has invested heavily in the cars’ processing power through Tensor Processing Units (TPUs), which are optimized for matrix multiplication and video processing. Furthermore, the TPUs work alongside NVIDIA’s GPUs and Intel’s CPUs to blend raw computing power with efficient machine learning.
In order to maintain the highest vision standards and to ensure lower costs, Waymo designs and manufactures its sensor suite in-house. The sensor suite seamlessly integrates Lidar, which can detect objects up to 300 meters away and create detailed maps of immediate surroundings; Radar, which can penetrate obstructions like rain, fog, or other cars; and advanced cameras, which are used for high-resolution color imagery.
Lastly, Waymo is able to maintain safety through its state-of-the-art prediction system, which is powered by VectorNet. VectorNet, a Deep Learning system, is built on graph neural networks (GNNs), which model each vehicle as a node in a weighted graph. The neural network is then able to predict the movement of surrounding vehicles through the use of real-world data.
Challenges
Since its inception, Waymo has faced a variety of regulations and public hurdles. During the recent Los Angeles Riots against increased ICE presence, several of Waymo's vehicles in Los Angeles were set ablaze by arsonists. As displayed in a variety of images circulating on the internet, the autonomous vehicles were further vandalized with a multitude of incendiary statements. This is not the first instance of public pushback—last year, a crowd in San Francisco set a Waymo aflame during a Lunar New Year celebration amid public distrust in autonomous vehicles.
Public distrust has not only fueled acts of protest but also justified the strict regulatory barriers that companies like Waymo continue to face. Waymo had to go through strenuous testing, development, and approval procedures to operate in the few cities that it does now. Furthermore, Waymo has faced extensive city-level resistance, especially from emergency responders, who have said that the vehicles are not well enough equipped for situations in which emergency vehicles need to get through. In conjunction with a reported inability to react to emergencies, many critics have raised concerns over edge cases, which include situations like construction zones, the use of hand gestures, and unpredictable pedestrian behavior. Additionally, the current lack of federal regulation of autonomous vehicles leaves Waymo in an uncertain position. As federal, state, and local governments begin to formulate policy targeting autonomous vehicles, Alphabet Inc. may find itself not only adapting to regulation but actively shaping it. With billions of dollars sunk into Waymo and deep influence in technology and policy circles, Alphabet is unlikely to walk away from its decade-long investment. Instead, the company may double down, leveraging its unique combination of data infrastructure, lobbying power, and powerful partnerships to push for standardized and clear regulation on the commercial use of driverless cars that favors well-established players, like itself.
While Waymo is still early in its early development and roll-out, there are monumental barriers to entry in the autonomous vehicle space that have left Waymo unprofitable. Currently, Waymo’s fifth-generation Jaguar I-Pace costs in excess of $200 thousand to produce the costly autonomy hardware. In addition to the cost of the vehicle, Waymo has invested heavily in technicians, charging and storage facilities, and regulatory staff—costs that are sure to multiply as it expands.
Implications
If successful in integrating their technology into society, Waymo could have a variety of wide-sweeping effects on urban environments. First and foremost, a widespread roll-out of autonomous vehicles would necessitate a change in traffic policy and emergency protocols. Furthermore, Waymo collects Lidar data as well as 360-degree video at all times, raising privacy concerns that would only be exacerbated by further integration into society. Lastly, if Waymo were to gain a high level of commercial traction in urban areas, many rideshare drivers could be left in the dust.
Waymo’s innovation in the autonomous vehicle sector displays new ways machine learning may enter our everyday lives; however, its proliferation could certainly face violent public pushback, cause logistic dilemmas, and displace rideshare and delivery drivers, raising concerns about income inequality and urban labor.