Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Explore the energy implications of autonomous vehicles as they evolve from prototypes to commercial deployments in this conference talk by transportation research software developer Haitam Laarabi from Lawrence Berkeley National Laboratory. Examine how autonomous vehicles are transforming into "Data Centers on Wheels" with substantial computing demands that could consume energy equivalent to all current data centers if 1 billion AVs operate just one hour daily. Analyze the three-tier framework covering ADAS systems (Level 2-3) with 30-100W onboard compute generating terabytes of data daily, current Level 4 autonomous vehicles featuring modular AI with GPU and FPGA-ASIC accelerators drawing 300-700W while producing hundreds of terabytes daily, and future Level 4+ vehicles using end-to-end foundational models potentially exceeding 1kW power consumption. Understand the challenges in quantifying energy footprints across onboard computing, data transfer, and datacenter training due to varied architectures and limited available data. Investigate how onboard computing can reduce vehicle range by up to 12% and assess the potential societal net energy savings in the transportation sector as these technologies see increased adoption.
