Toronto-based Waabi, a company specializing in autonomous truck technology, announced on Tuesday a strategic collaboration with Volvo Autonomous Solutions. This partnership focuses on the joint development and deployment of self-driving trucks. The agreement brings together Waabi’s advancements in generative AI with Volvo’s autonomous truck, the Volvo VNL Autonomous.
A key aspect of this collaboration is the integration of Waabi’s virtual driver system, the Waabi Driver, into Volvo’s autonomous vehicle, which is designed with redundant systems to ensure safe operation.
Speaking with FreightWaves, Waabi’s founder and CEO, Raquel Urtasun, clarified that this is not a retrofitting process but a vertical integration. The trucks will be factory-built with all necessary autonomous technology already installed.
“We don’t see retrofitting as a viable option for a redundant platform. Our approach, and that of any AV company, should be a vertical integration into a platform specifically designed for self-driving.” Urtasun stated.
While Volvo is responsible for manufacturing the truck itself, Waabi has been developing the Waabi Driver, which is part of a new phase in autonomous vehicle evolution known as AV 2.0.
AV 2.0 and the Evolution of Autonomous Trucking
To fully grasp AV 2.0 and its impact on autonomous trucking, it helps to first understand its predecessor, AV 1.0.
“There was an earlier generation of technology in this industry, referred to as AV 1.0. This approach relied on engineered systems where AI played a minor role” Urtasun explained.
She likened AV 1.0 to a network of smaller AI models that required extensive human oversight. Large teams had to program the system and conduct vast amounts of real-world testing to address challenges it couldn’t handle. This method required significant time and financial investment, making progress slow and resource-intensive.
In contrast, the past two years have seen the emergence of AV 2.0, which allows a single AI system to handle all driving-related tasks. Urtasun drew a parallel to large language models (LLMs), where AI can now design systems that previously had to be manually programmed.
This new level of AI capability demands significant computing power. Companies like Nvidia have been making headlines in the trucking sector because their hardware is driving the AV 2.0 transformation. The Nvidia Drive Thor platform, built on the Blackwell architecture, is designed to support AI workloads like transformers, LLMs, and generative AI. This same platform also powers the Waabi Driver.
A downside to AV 2.0 is its “black box” nature. While it enables AI to function independently, verifying that the system operates correctly can be challenging. The approach requires vast amounts of data, shifting the burden from human effort to data centers and computing power, leading to substantial costs.
The recent breakthroughs in AI, such as those from Deepseek, have attracted attention because they offer solutions to this data-intensive problem. By improving AI’s reasoning capabilities, these advancements help strike a balance between data efficiency and computational demands.
Waabi World – A Virtual Training Ground for AI
Advancements in AI for autonomous driving increasingly resemble the creation of a virtual human brain. Traditional AV 1.0 methods required enormous human effort for manual programming, while AV 2.0 demands immense computational resources to process and learn from real-world data. This presents an ongoing challenge in AI development.
To overcome these obstacles, the next logical step was to build an AI-driven training environment for self-driving vehicles. Relying solely on real-world driving tests would require an enormous fleet of autonomous trucks driving millions of miles to encounter and learn from every possible road scenario.
The Waabi Driver functions as the AI brain behind self-driving technology, but it requires a high-fidelity simulator to train effectively. Waabi World, a generative AI-powered virtual environment, addresses this need by offering four key capabilities. It creates digital replicas of the real world at scale, conducts high-precision sensor simulations, allows immersive testing, and enables stress testing to refine the Waabi Driver’s abilities.
By training in this virtual space, the Waabi Driver can learn from simulated mistakes and refine its skills before ever hitting the road.
A compelling example of this system’s effectiveness came when the Waabi Driver first encountered rain. Despite never having been trained for rainy conditions, it adapted seamlessly and continued driving without issue.
An added advantage of Waabi’s approach is the symbiotic relationship between the AI simulator and the Waabi Driver. Urtasun explained, “One acts as the teacher, and the other as the student. Through continuous learning, they improve each other over time, eliminating the need to wait for real-world conditions to arise.”
As Waabi continues to push the boundaries of AV technology, the company has already laid the foundation for integrating the Waabi Driver into the Volvo VNL Autonomous. Testing is expected to begin in 2025, according to a joint press release.
This partnership builds upon Volvo’s long-term commitment to autonomous vehicle development. In January 2023, Volvo Group Venture Capital became a strategic investor in Waabi and later participated in the company’s $200 million Series B funding round.
With these advancements, Waabi and Volvo are poised to shape the future of autonomous trucking, setting new industry standards for AI-driven transportation.
Volvo, a global leader in commercial vehicle manufacturing, has consistently pushed the boundaries of innovation in trucking. By teaming up with Waabi, a startup renowned for its cutting-edge AI simulation technology, Volvo is reinforcing its commitment to shaping the future of autonomous freight transport. The partnership leverages Waabi’s advanced AI technology alongside Volvo’s expertise in vehicle manufacturing to create more reliable and efficient self-driving trucks.
