Kodiak Targets Class 8 Electric Truck for Autonomous System
The Peterbilt Model 579EV employs Kodiak Robotics’ fifth-generation sensor stack that relocates new SensorPods to the side-mounted mirrors.
Kodiak Robotics introduced what it claims is an industry-first at the 2023 Advanced Clean Transportation (ACT) Expo: an autonomous Class 8 truck that is fully electric. Kodiak upfitted a Peterbilt Model 579EV electric truck with its latest SAE Level 4 automated-driving system, the Kodiak Driver.
“It is the first-ever autonomous electric truck, not only for Kodiak but for the industry,” Michael Wiesinger, VP of commercialization for the five-year-old autonomous-tech startup, told SAE Media during a vehicle walkaround in Anaheim, California.
The 579EV is the second vehicle platform that Kodiak has upfitted with its fifth-generation sensor stack that was introduced in April 2023. The new sensor suite already is integrated on Kenworth T680 diesel-powered trucks in Kodiak’s fleet. The integration on the 579EV, which will join Kodiak’s fleet in 2024, illustrates the company’s assertion that its autonomous system is vehicle- and powertrain-agnostic. Wiesinger said that Kodiak will explore other sustainable vehicle platforms, such as hydrogen fuel-cell trucks, as they are introduced to the market.
The Peterbilt Model 579EV can be recharged in three hours and provides a peak 670 hp (500 kW). Designed for short-haul and drayage deployments, the truck has a range of up to 150 miles (241 km). Wiesinger said that as EV technology advances and range increases, Kodiak will be well-positioned to integrate its autonomous stack into future EV platforms. Studies have demonstrated up to a 10% reduction in fuel consumption by using automated-driving technology, he noted – which for electric trucks translates into often-critical extended driving range.
Bye-bye ‘unibrow’
The fifth-generation autonomous-truck platform increases sensor redundancy and GPU processing power. And noticeably, the roof-mounted “center pod” sensor suite is gone. Instead, Kodiak opted to relocate the front-facing Luminar Iris lidar and wide field-of-view camera to proprietary mirror-mounted SensorPods on each side of the truck.
“The most important thing is it doesn’t have anything on the roof anymore. What others call the ‘unibrow,’ we’ve shaved it off,” Wiesinger said. “When you talk to fleets they say, ‘This [mirror module] is one structure that I can easily take off and put on again.’ [The location] really resonates with them from a maintainability and an uptime perspective.” A SensorPod can be replaced in as little as ten minutes, he added.
Enhanced perception capability is another benefit; the sensors are placed at a human driver’s line of sight. The new location reportedly doubles the Kodiak Driver’s lidar coverage at long range. It also simplifies the build process, Weisinger said.
The latest-generation Kodiak Driver adds a second forward-facing lidar and additional camera compared to the previous center-pod setup. “Before, we had sensors in the middle [of the vehicle] – now we have them left and right, so there’s additional redundancy from a sensor perspective,” Wiesinger said. “When we designed the SensorPods, we did a field-of-view analysis specific for the long-haul-trucking use case and have identified the best setup.”
Fewer required modifications and touch points enable more cost-effective integration into trucks, which marks a significant step toward commercial deployment of the Kodiak Driver, he said. “In the future, if you think about potentially moving to an [camera-based] e-mirror – a concept that some truck manufacturers are exploring – then we can take away the glass [mirror] and make the SensorPod smaller,” Wiesinger said. “We already have thoughts and designs of what that next generation could look like.”
Robust sensor array
The fifth-generation Kodiak Driver platform as demonstrated on the Kenworth T680 increases the total number of onboard sensors from 14 to 18, including one new lidar and three new cameras – bringing the total camera count to 10. Two wide-angle cameras (not shown on the 579EV) were added to the hood-mounted mirrors to cover blindspots.
Kodiak’s long-range sensor suite includes the following, split evenly between the two side-mounted SensorPods: four ZF Full Range Radars; two Hesai 360-degree scanning lidars for side- and rear-view detection; two Luminar Iris lidar sensors; and 8 cameras that include both wide and narrow field-of-views. The new application also features the recently-released Ambarella CV2 perception system-on-chip (SoC) that handles all camera data processing. The Ambarella CV2 SoC is said to improve image quality for longer-range detections and enhance dynamic range for nighttime driving.
“We are providing the trucking industry with the processors it requires to commercialize higher levels of autonomy,” Fermi Wang, president and CEO of Ambarella, said in a statement earlier this year. “With Kodiak as a partner, we’re confident in our ability to serve this vital segment [long-haul trucks] of the emerging self-driving industry.”
Kodiak’s fifth-generation truck also includes aerospace- and military-grade technology, according to Wiesinger. The main connector for data and power in the SensorPods and compute system was designed for use in aviation, aerospace and military applications. The compute hardware is manufactured by Crystal Group.
Engineers improved the hardware stack – which includes two onboard custom-designed safety computers, the Kodiak Actuation Control Engine – and reduced the electrical power requirements while improving the processing power of the system. In addition to 130% more GPU processing power, the new system reportedly provides 60% more central-processing capability and additional system redundancy. The reduced power consumption allowed for a 50% reduction in the size of the fifth-generation power system and decreased cooling needs.
On the defense
Kodiak Driver-equipped trucks deliver freight daily for customers such as IKEA and Tyson Foods across the southern U.S., operating autonomously on the highway portions of the routes. Wiesinger said its trucks have accumulated more than 115,000 miles (185,000 km) of safe autonomous operation for IKEA, running between Houston and Dallas.
“We are running our trucks on the road, hauling freight for our customers. But in the future, we want to integrate our system onto their trucks. That’s our future business model,” Wiesinger said, noting that longer routes such as Dallas to Atlanta are ideal for autonomous trucks. The startup currently focuses solely on the U.S. market, but it “definitely has global aspirations,” he added.
The company now is laying the groundwork for future expansion beyond long-haul trucking. In December 2022, Kodiak was awarded a $49.9 million, 24-month U.S. Department of Defense agreement to help automate future U.S. Army ground vehicles led by the Army’s Robotic Combat Vehicle (RCV) program. The vehicles will perform reconnaissance, surveillance and other high-risk missions.
Kodiak says it was the only autonomous-vehicle company initially selected for this award out of 33 submissions. The first year will be devoted to applying its autonomous software stack for Army-specific use cases. In year two, the company will implement its autonomous-driving system on off-road vehicles capable of operating remotely in complex, unpredictable conditions.
Top Stories
INSIDERDesign
This Robot Dog Detects Nuclear Material and Chemical Weapons
ArticlesEnergy
INSIDERDesign
New Anduril, Skydio Drones Start Field Testing in Romania
INSIDERData Acquisition
Atomic Fountain to Host Defense Research for Quantum Sensing
INSIDERAerospace
Testing the Viability of Autonomous Laser Welding in Space
INSIDERRF & Microwave Electronics
Webcasts
Power
Designing an HVAC Modeling Workflow for Cabin Energy Management and XiL Testing
Unmanned Systems
Countering the Evolving Challenge of Integrating UAS Into...
Aerospace
Best Practices for Developing Safe and Secure Modular Software
Manufacturing & Prototyping
How Pratt & Whitney Uses a Robot to Help Build Jet Engines
Unmanned Systems
Scaling Manufacturing and Production for 'Data as a Service' Electric Drone
Test & Measurement
A Quick Guide to Multi-Axis Simulation and Component Testing