Waabi World Is a Driving School for Autonomous Trucks

The trucking industry is the backbone of the modern economy. If all trucks stopped hauling for a day, it would be a considerable shock to life as we know it.

Truckers work very long hours on a job that demands alertness. These hours are spent alone, away from home and family, for weeks or even months. Due to the lifestyle demands and relatively low pay for hours worked, there is a shortage of willing long-distance truck drivers in North America and Europe.

Trucking also suffers from a negative public perception as a low-status occupation. A large percentage of current truckers are over 55 years old, and most will retire over the next five years.

Several companies are pushing autonomous trucks to fill the supply gap, and a Toronto-based startup, Waabi, is one of them.

Waabi says it is building generative AI applications for the physical world, starting with self-driving, long-haul trucks. It was founded by Raquel Urtasun, former chief scientist of Uber's autonomous vehicle unit, in 2021.

Autonomous freight is a challenging problem to solve, but it promises many benefits, such as new scheduling options, faster delivery windows, and fewer accidents involving delivery trucks.

Waabi has created two generative AI systems to revolutionise self-driving: Waabi World, a neural simulation engine, and Waabi Driver, a commercial autonomous truck that has been operating on public roads since 2023. The company trains its trucks using mixed-reality testing, a method that combines real environments with synthetic scenarios generated by Waabi World.

Waabi World is described as the “ultimate school” for self-driving vehicles. It teaches the Waabi Driver to drive autonomously by exposing the car to various scenarios, including common situations and edge cases. The company says it reduces the need to drive testing miles in the real world and makes testing cheaper and safer. On-road driving is saved for final validation and verification.

The Waabi Driver combines an onboard version of the Waabi simulator with LIDARs, cameras, and radars, and responds to modified sensor data flows created by the simulator. This training approach allows the truck to generalize to all possible situations on the road, including those it has never encountered.

Waabi is starting with autonomous trucks, but it plans to rapidly and safely commercialize other self-driving vehicles. The company’s fleet operates on commercial routes with carriers and shippers across Texas.

There will be some challenges on the road to widespread adoption of Waabi’s vehicles. A major challenge to self-driving vehicles is accountability. If an autonomous truck is involved in a collision, who should be held responsible: the manufacturer or the hauling company?

Some people are concerned that autonomous trucking may mean fewer jobs for human drivers. However, a truck driver’s day-to-day routine involves maintenance, inspections, loading/unloading, fueling, and protecting cargo — all complex tasks that are harder to automate than highway driving.

Other companies are in the race to develop self-driving trucks, including Aurora, Kodiak, Plus, Daimler Truck, Waymo, among others. Fully autonomous trucks are expected to be on highways by 2027.

Cabe Atwell previously wrote about self-driving cars that could “recall” the way to familiar destinations.