Torc Robotics and Daimler Trucks North America’s co-development of autonomous truck technology takes the next step, as the second iteration of test trucks designed to work with Torc’s SAE Level 4 autonomous self-driving technology hit the road for testing.
The two companies have been testing their technology near Torc’s Virginia headquarters, but are expanding their testing into the Southwest in early 2021, using an enhanced prototype truck. This new generation of Freightliner Cascadia test trucks, internally known as Gen 2, are the second iteration of trucks jointly developed by the two companies.
Torc and Daimler Trucks’ ultimate goal is to “reinvent the truck” by co-developing a Level 4 Freightliner Cascadia that includes safety-critical redundancy components, as well as the seamless integration of additional computers and hardware required for self-driving technology.
“To meet the redundancy and performance requirements of a self-driving truck, the traditional truck chassis must be reinvented,” explained Michael Fleming, Torc’s CEO. “Just like any major innovation, it requires a stair-step approach toward the final product. We are taking this one step at a time, with safety as our guiding principle.”
Torc has been commercializing its self-driving technology in heavy-duty applications for more than a dozen years. In 2019, Torc and Daimler Trucks teamed up to bring self-driving trucks to market within the decade.
“We knew from the outset that self-driving technology cannot be commercialized without an OEM,” Fleming said.
“Our partnership with Torc is critical to our efforts to commercialize a Level 4 highly automated truck,” said Roger Nielsen, president and CEO of DTNA. “Torc’s experience with developing self-driving technology and their focus on safety makes them the ideal partner. Our joint goal is a Level 4 integrated truck that provides true customer value.”
Level 4 Autonomous Trucks
Level 4 autonomous control, as defined by the Society of Automotive Engineers, allows for complete driver inattention to vehicle activity, with no no driver attention ever required for safety. The driver may safely go to sleep or leave the driver's seat and only need resume control of the vehicle in areas with limited telemetry or under special circumstances, such as accidents or traffic jams. Outside of these areas or circumstances, the vehicle must be able to safely abort the trip and safety park the vehicle if the driver does not retake control.
Market-ready self-driving trucks must emulate the actions of the most experienced and safe truck drivers, say Torc and Daimler. The partners are developing software and hardware that is seamlessly integrated to reliably handle failures of safety-critical vehicle components, such as braking, steering, power distribution, and messaging.
The team’s vision for a Level 4 vehicle platform is one in which component redundancies and software behaviors work together. In the case of a brake failure in a Level 4 truck, for instance, redundancies would maintain the vehicle’s ability to decelerate and stop without human intervention. Torc’s self-driving software would then be able to maneuver to a safe location so a support crew could service the brake system, according to Fleming.
Another integral behavior the team is working to replicate is the way experienced truck drivers are able to feel component failures. “Our software engineers are working with highly skilled truck drivers to understand this experience and transition this human intuition into embedded sensors and algorithms,” Fleming said.
Iterative Truck Generations for Self-Driving Tech Development
Torc and DTNA expect to develop multiple iterative test truck models before they release a self-driving truck for commercial customers. The prototypes will incorporate many lessons learned from testing and development since the partners started working together in 2019 and Daimler acquired a majority stake in the autonomous-technology company.
The upgrades included in the Gen 2 prototype truck are designed to bolster the testing effort and speed up the collection of data that is used in machine learning and algorithmic development.
Improvements in data collection and transfer will help the development team expand capabilities quickly and safely, according to the companies, supporting Torc’s effort to scale its test fleet in multiple locations. Additional sensor density and coverage will assist in overall high-fidelity perception performance and long-range sensing power, critical for highway driving.
Both companies have stated that they will only deploy self-driving trucks when they are safe and reliable – not by a set date. Fleming is convinced the team will meet its goal, because “We are two pioneers joining forces – we understand the complexities of commercializing self-driving technology. Our mission is to save lives and our vision is to become the standard in self-driving trucking.”
In a Zoom call with reporters last month, Peter Vaughan Schmidt, head of Daimler Truck’s Autonomous Technology Group, explained that the plan is to deploy small pilots with customers, “complete a few missions in selected [operations] in selected areas in the Southwest region of the U.S., and then grow it from there over time.
“It’s a marathon, it’s not a sprint. Safety will dictate our timeline."