|
Take heed to this text  |
Kodiak’s sixth-generation driverless truck. | Supply: Kodiak Robotics
Kodiak Robotics Inc. launched its sixth-generation, driverless-ready semi truck at CES in Las Vegas this week. The corporate mentioned its self-driving truck is designed for scaled deployment and builds on 5 years of real-world testing. This testing included 5,000 masses carried over greater than 2.5 million miles.
The Mountain View, Calif.-based firm mentioned it would use the brand new truck for its personal driverless operations on roadways between Dallas and Houston this yr.
“Over the course of two.5 million miles, we’ve efficiently demonstrated that our self-driving vehicles can stand up to the cruel atmosphere of long-haul trucking from each a platform integrity and a software program perspective,” mentioned Don Burnette, founder and CEO of Kodiak, in a launch.
“This truck basically demonstrates that we’ve achieved the work vital to soundly deal with driverless operations,” he added. “Whereas we proceed to work with main truck producers, the expertise we developed is deployment-ready, uncoupled from OEM timelines and truck manufacturer-agnostic, which permits us to maneuver quick whereas holding security on the forefront.”
Kodiak mentioned the newest version of its autonomous semi-trucks has all the required redundant safety-critical {hardware}. These embrace braking, steering, and sensors, in addition to the software program required for driverless operations at scale.
Redundancy is the important thing to security, says Kodiak
Kodiak famous that its newest autonomous semi-truck consists of redundancy throughout all safety-critical capabilities. These embrace a redundant braking system, redundant steering, and redundant energy.
Whereas conventional semi-trailer vehicles have redundant braking techniques, Kodiak mentioned it wished to take this additional. Its vehicles’ pneumatic braking techniques include three particular person brake actuators.
All three actuators are concurrently managed by Kodiak’s system. Which means if any of them fail, the backup techniques can forestall lack of management and produce the truck to a protected cease.
The corporate has additionally geared up its vehicles with a dual-redundant steering system that features two redundant ZF actuators managed by Kodiak’s security system.
Primarily based on the firm‘s security evaluation, if the first steering actuator experiences any kind of failure, then the steering system switches to the secondary actuator. This enables the system to keep up full management and transfer the car right into a protected state with out compromising the car dynamics.
The vehicles have a redundant energy system that powers the computer systems, sensors, actuators, and all different electrical techniques. Kodiak mentioned its energy system is break up into two totally remoted subsystems. These subsystems make sure that all security techniques can nonetheless carry out a protected “fallback” ought to both fail.
As well as, Kodiak asserted that each one of its vehicles, not simply the latest era, are geared up with its custom-designed Actuation Management Engine (ACE). The ACE system is a high-integrity security pc answerable for making certain that the Kodiak Driver can information the truck to a fallback out of the circulate of site visitors ought to a crucial system failure happen.
Lastly, the truck features a redundant LTE communications hyperlinks, permitting the truck to ascertain dependable communications with Kodiak’s redundant command facilities in Lancaster, Texas, and Mountain View, Calif.
Kodiak’s newest era of vehicles options upgraded SensorPods and extra. | Supply: Kodiak Robotics
Contained in the sixth-generation truck
Like its earlier vehicles, Kodiak’s sixth-generation car options its Kodiak Driver, a vehicle-agnostic self-driving system and driverless-ready {hardware} platform. The vehicle-agnostic elements of the driving force will come in useful as a result of the corporate plans to roll out its sixth-generation platform to a number of car sorts.
Kodiak claimed that its newest vehicles function twice the GPU processor cores, 1.6 occasions better processing pace, 3 times extra reminiscence, and a pair of.75 occasions better bandwidth to run software program processes than its first-generation truck.
This newest era of vehicles additionally options upgrades to Kodiak’s proprietary SensorPods, which home the sensors and are pre-calibrated and pre-built for quick and straightforward repairs. Kodiak improved its SensorPods with two higher-resolution lidars, which at the moment are automotive-grade. It additionally has two further facet radar sensors to enhance long-range detection.
In complete, Kodiak geared up its vehicles with 12 cameras, 4 lidar sensors, and 6 radar sensors. To course of this elevated sensor knowledge, the self-driving truck maker depends on NVIDIA GPUs.
The corporate additionally added top-mounted, extra-bright hazard lights to its SensorPods. Kodiak designed these lights to adjust to the autonomous trucking business’s utility for an exemption to Federal Motor Car Security Regulation 392.22. This regulation requires conventional truck drivers to put warning units on a roadway after a breakdown.
Since driverless vehicles gained’t be capable of place highway flares or different units alongside the roadway, Kodiak mentioned these lights shall be used to alert different drivers to the presence of a truck on the facet of the highway, pending Federal Motor Service Security Administration approval.
Kodiak additionally geared up its vehicles with microphones, in order that its vehicles can detect and determine the presence of emergency automobiles and different suspicious sounds that might symbolize a hazard.
Kodiak’s sixth-generation truck follows the corporate’s partnership with Pilot to open a truckport for launching and touchdown autonomous vehicles. It has additionally teamed up with Loadsmith, C.R. England and Tyson, Ahead and extra.
Kodiak plans to proceed enhancing its self-driving vehicles. As an example, it plans to combine an Ambarella CV3-AD AI area management system on chip (SoC).
