The journey from paper notepads to driverless trucks has unfolded faster than most in the industry anticipated. Decades of technological evolution compressed into roughly 15 years of intense innovation. Today, as embedded computing catches up to artificial intelligence ambitions, autonomous trucking stands at an inflection point where software optimization matters more than raw computing power.
From paper logs to multimodal sensors
“When it comes to truck technology, we’ve come a long way,” said Rehan Tahir, who works on embedded automotive computing at AMD, in a FreightWaves interview. “In the beginning everything was being logged on paper. Notepads. And then they started to do that digitally.”
This digital transition opened the door for driver monitoring cameras, forward-facing stereo vision systems, and eventually the multimodal sensor arrays that power today’s Level 4 autonomous trucks. Tahir noted that AMD has been deeply involved in forward camera technology “for the good part of a decade,” developing specialized algorithms for both monocular and stereo vision applications.
The hardware scaling challenge
Early autonomous truck developers faced a major hurdle. They crammed bulky data center components into sleeper bunks, creating systems that were too costly, too power-hungry and too unwieldy to scale.
“A few years ago some of these trucks had data center components sitting in the back,” Tahir said. Over time those server farms shrank to the size of college dorm refrigerators.
“Until we had the kinds of capability in an embedded platform, you could really start integrating,” Tahir said. “We’re going from single-digit TOPS to hundreds, sometimes thousands of TOPS.”
Modern Level 4 systems can run more than 20 cameras alongside four to six lidar units and a similar number of radars. Fusing all that data often creates what one AMD customer called “almost a bigger compute problem than doing the actual AI.”
Automotive-grade standards
Automotive chips must meet strict requirements. These include AEC-Q100 standards for temperature durability and ASIL B or ASIL D functional safety. Regulators such as the National Highway Traffic Safety Administration (NHTSA) and New Car Assessment Program (NCAP) will not certify systems built on consumer-grade components.
“Yes, they come to us and they say, guys, we cannot scale this server farm,” Tahir said. “Number one, it’s too costly. Number two, it’s too power-hungry. Number three, it’s not gonna meet the safety and security regulatory requirements.”
AMD has shipped more than 300 million devices into the automotive space over two decades. More than half of those went into Advanced Driver Assistance Systems (ADAS) and autonomous driving applications.
AI momentum and hard lessons
The explosive arrival of ChatGPT accelerated the industry’s understanding of AI’s true potential. Vision-language-action models now power perception systems in autonomous trucks. Tahir expects agentic AI experiences to become standard inside vehicles, just as they have on phones and PCs.
Companies also learned difficult lessons about moving too fast. Tahir pointed to the major consolidation wave that swept through robotaxi and autonomous trucking companies around 2022. Many firms had impressive proof-of-concept systems but lacked viable business models.
“It’s easier to go to the moon than it is to build an autonomous truck,” Tahir said. “Everyone understands how challenging this is.”
The cabless future
Purpose-built designs that eliminate the cab entirely represent the next major frontier. Zoox already operates steering-wheel-free robotaxis in Las Vegas. Tahir believes autonomous trucking will follow a similar path once public comfort levels rise, although he cautioned against rushing the transition.
“You do these things too quickly and you end up getting a backlash,” he said. “You don’t want to be the one that was like, oh yeah, we’re moving so fast and not being responsible about it.”
Freight Fraud Symposium
Double brokering. AI deepfakes. Identity theft. Freight fraud is an existential threat to the industry. Get ahead of it.
Supply Chain AI Symposium
Past the hype. Join operators, founders, and enterprise leaders figuring out how to deploy AI in supply chain.
Future of Rail Symposium
Reshoring is rewriting freight demand. Join shippers, rail executives, and government officials to shape the next decade.
Double brokering. AI deepfakes. Identity theft. Freight fraud is an existential threat to the industry. Get ahead of it.
Rock & Roll Hall of Fame • Cleveland, OH Register NowPast the hype. Join operators, founders, and enterprise leaders figuring out how to deploy AI in supply chain.
The Old Post Office • Chicago, IL Register NowReshoring is rewriting freight demand. Join shippers, rail executives, and government officials to shape the next decade.
The Signal at Chattanooga Choo Choo • Chattanooga, TN Register Now