Are Self-driving Cars Driving Global Carbon Emissions in the Wrong Direction?

by Carolyn Mathas

MIT researchers found that 1 billion autonomous vehicles on the road just one hour per day with a computer consuming 840 watts, will consume enough energy to generate emissions on par with today’s data centers.

A new study from MIT researchers explored the potential energy consumption and related carbon emissions should wide adoption of these vehicles occur.

Data centers are widely known for their large carbon footprint: accounting for 0.3 percent of global greenhouse gas emissions. MIT researchers built a statistical model to study the problem. They found that in over 90 percent of scenarios, to keep autonomous vehicle emissions from passing data center emissions, each vehicle must use less than 1.2 kilowatts of power for computing, which means they’ll need more efficient hardware. If, for example, 95 percent of the global fleet of vehicles is autonomous in 2050, given the resulting computational workloads, hardware efficiency would need to double faster than every 1.1 years to keep emissions under control.

The researchers built a framework to explore the operational emissions from computers on board a global fleet (no human drivers). They also needed to model advanced computing hardware and software that doesn’t exist yet. They modeled the workload of a popular algorithm for autonomous vehicles, known as a multitask deep neural network. Then the team explored how much energy the network would consume if it simultaneously processed many high-resolution inputs from multiple cameras with high frame rates.

Autonomous vehicles can move goods and people, so they could distribute massive amounts of computing power along global supply chains. And their model only considers computing — it doesn’t consider energy consumed by vehicle sensors or the emissions generated during manufacturing.

One way to boost computing efficiency could be specialized hardware designed to run specific driving algorithms. However, vehicles tend to have 10- or 20-year lifespans, so a challenge is to “future-proof” the specialized hardware to run new algorithms. Algorithms can also be more efficient, requiring less computing power. However, trading off accuracy for more efficiency could hamper vehicle safety.

Related Articles

Leave a Comment

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Privacy & Cookies Policy