By now, autonomous vehicles are anything but science fiction. Though readiness is a mixed bag, General Motors (GM) has established an ambitious autonomous development pipeline. CEO Mary Barra used CES 2022 to highlight GM’s approach—claiming that personal autonomous vehicles (AVs) will be hitting production lines “as soon as mid-decade." 

GM hopes to strengthen its foothold on the EV market with rollouts like the Chevrolet Equinox EV SUV. Image used courtesy of GM

Though this wording affords GM some wiggle room, Barra hopes that average buyers can experience autonomous driving both efficiently and safely. A partnership with Cruise—an emerging autonomous ride-share service—aims to accelerate that launch. The duo may also build autonomous networks needed to enable public transportation services.

What obstacles lie in the way? 

EVs Slowly Rise in Popularity 

Naturally, claims of full autonomy will draw plenty of skepticism. Both members of the general public and the automotive world question if technology can adequately mature in the coming years. Electric vehicle (EV) uptake is also relatively slow versus ICE technology—despite the growing popularity of EVs in 2021. Although all-electric car sales nearly doubled from 2020, they still account for a mere 4% of total domestic sales across all segments. 

The annual development of EV fleets between 2015 and 2020. Image used courtesy of the Guardian

However, electric vehicles made up 9% of total car sales in China—a country where GM enjoys massive popularity, and in which the company sold the most overall units. GM’s planned introduction of 30 electric vehicles globally by 2025 will be crucial. Fortunately, this signals serious investment and research into battery power. Demand will determine how seriously GM takes EV development. 

GM Works Against the Court of Public Opinion

While there’s a buzz surrounding AVs, everyday drivers won’t embrace the technology so readily. Consider the following survey results from the past few years—pulled from a variety of sources: 

• According to a 2019 AAA poll, 71% of U.S. drivers would feel unsafe riding in a fully-autonomous vehicle.
• 67% of respondents from a 2019 Reuters-Ipsos poll believe that self-driving cars should be held to higher safety standards than existing cars.
• 57% of Americans cite lack of interest in self-driving vehicles—much of that apathy stemming from safety concerns—according to a 2018 Allianz survey.
• 84% of buyers prefer that manual operation remain an option at all times, even within autonomous vehicles, according to a 2018 Cox Automotive survey.

What can we learn from these opinions? America as a whole still possesses plenty of distrust in autonomous technology. Brands like GM must convince the public that self-driving cars can better their everyday lives. 

While it’s tempting to write off GM’s autonomous vehicle as a proof of concept, the automaker itself would disagree. GM does have success on the EV front, stemming primarily from its Bolt model. Additionally, its experience with battery propulsion has continually evolved following the 1996 launch of the EV1. The R&D team isn’t starting from square one. That said, what inroads are being made? 

GM's EV1. Image used courtesy of GM and Car and Driver

How is GM Working Toward Autonomy? First, EV Batteries

Let’s start with the heart of the AV: the battery. GM has long outfitted its EVs with lithium-ion (Li-ion) batteries, and other automakers have followed suit. Though alternative compounds like lithium-sulfur, sodium-ion, zinc-ion, and aluminum-air setups have promise, they’re far from ubiquitous. Until a new technology attains large-scale viability, manufacturers will refine their lithium-ion units. 

The same is true for GM, which has recently unveiled its new Ultium battery technology. Ultium relies on a pouch-type Li-ion cell, and these cells together form modular battery packs. Based on a nickel-cobalt-manganese-aluminum chemistry, these units are easy to scale up or down to match the desired vehicular platform. That means larger vehicles can utilize a larger kWh pack, while smaller cars might incorporate a pack with less power. GM’s Ultium battery ranges anywhere from 50 kWh to 200 kWh in capacity.

GM's Ultium battery pouch cells employ a modular design, unlike cylindrical battery cells. Image used courtesy of GM

The battery technology features the following: 

• Stackable cells both vertically (like slices of bread) and horizontally (like pancakes)
• Embedded intra-module electrical components, which means 80% battery pack wiring
• Smaller form factors for design flexibility and passenger comfort
• Potential weight savings, better aerodynamics, and improved aesthetic
• Options for running battery layers in parallel or in a series 
• Massive savings with costs approaching just $100 per kWh

Ultium packs alone cost 90% less than the 2010 Volt’s battery packs once did. Fast DC charging doesn’t erode usable capacity over time. Finally, GM expects each pack to last anywhere from 150,000 to 250,000 miles in real-world conditions. That outlasts the Bolt—which has been highly successful for GM whilst remaining approachable for buyers. 

Onboard Sensors and Cameras

Sensors will also be essential components in GM's plan for full autonomy. This is where Cruise’s proprietary AV sensors come into play. LiDAR has been indispensable when it comes to mapping vehicular surroundings in real-time—tracking scenery, potential obstacles, other vehicles, and pedestrians. The LiDAR sensor uses an array of light pulses to take these readings, then collects any returning pulses. 

Unfortunately, the cost of these units has been astronomical in many instances. Through its partnership with Cruise and GM's acquisition of Strobe, GM is able to source these units at roughly $300 apiece.

Strobe, which was snapped up by GM and Cruise, uses LiDAR-on-a-chip technology. Image used courtesy of GM and Medium

That’s a far cry from units that can cost up to $20,000. Even though GM plans to use five redundant LiDAR units per AV, their small cost and compact sizing won’t challenge engineers. 

Otherwise, GM’s upcoming AVs may feature the following components, should they closely mirror Cruise’s models: 

• 14 cameras to detect traffic lights, spacing, human beings, and other surroundings
• 3 articulating radar units within the car’s mirrors to provide wide-field coverage at long ranges for cornering, parking, and navigating intersections
• 8 front-facing radar units to track distances and speeds of other vehicles
• 10 front-facing, short-range radar units to sense nearby objects

Software and Analysis

These hardware components are matched to software, which interprets the data collected constantly from the vehicle. This helps the car react to its surroundings and operate effectively. Baked-in machine learning allows for testing and simulations. These help each AV distinguish between traffic signals, detect orientation, analyze visuals, and optimize LiDAR functionality. 

The brain of the car will reside within the engine control module (ECM), which controls a number of vehicular functions. The chipset that allows all onboard electronics to work helps the car navigate the streets safely by ingesting and processing incoming data. This SoC setup will prove integral to GM’s own Ultra Cruise driver-assist technology, which will leverage an AI-capable 5nm processor, according to GM's 2021 investor presentation. Ultra Cruise is said to handle 95% of driving scenarios, which will come in handy for AVs. Like a conventional AV, this technology will also rely on cameras, LiDAR, and radar. 

Meanwhile, Super Cruise will provide a stripped-down version of this hands-free system—providing lane centering, adaptive cruise control, and emergency braking (on compatible roadways). 

A Tenuous Timeline for Barra's Lofty Goal

It appears that the pieces are in place for GM’s success, though the timeline is still up for debate. The company plans to bring Super Cruise and Ultra Cruise, for example, to multiple vehicles next year onward. These features are an interesting start, though they do rely on driver attention to work properly. It remains to be seen whether Cruise’s contributions can help GM take that next step towards complete autonomy. 

GM's timeline for EV and ICE production. Image used courtesy of GM

Cruise completed much of its live testing in 2020. With its test fleet completing over 770,000 miles of trials, the AV platform boasted impressive safety and reliability. This is typically measured via disengagements per 1,000 miles; in these situations, the driver is forced to take control or some critical technology failure occurs. Cruise recorded just 0.035 disengagements during this period. Though that measurement’s validity is debated, Cruise only recorded one disengagement per 28,520 miles on average in 2020. 

Cruise’s safety trajectory has been promising. Its vehicles performed ten times better in the final three months of 2020 than they did one year prior. The company claims that disengagements were largely prompted by errors from other drivers—not functional shortcomings. However, these controlled tests might not accurately reflect real-world performance. 

Messaging from GM and government officials will also be key to full-autonomy adoption. Consumers must get on board, and regulations impacting AVs will undoubtedly evolve in the next handful of years. Will GM meet Barra’s ambitious goals? That’s uncertain, though electrical engineers and designers will be central to overcoming fundamental challenges.