As Excitement for Autonomous Trucking Grows, Fleets Have Many Products to Choose From

While self-driving cars are in the news all the time, self-driving trucks often slip under the radar. This is interesting since both autonomous cars and trucks try to automate activity on the same roads. When we look at the big picture, the US trucking business is currently valued at $720 billion and employs almost a million people. Today we will look at the autonomous trucking industry and take a look at some of the most interesting developments taking place today, as well as the data annotation service required to make these technologies a reality. 

Aurora 

Aurora is a company that is partially built on its acquisition of Uber’s self-driving technology and calls its fleet of trucks Aurora Horizon. They are starting out in Texas and then hope to expand around the Sunbelt and later to other parts of the US as commercialization is expected to begin in late 2023. One of the interesting technologies they use is something called “FirstLight LiDAR,” which, according to Aurora, is better than traditional LiDAR because of its long-range sensing capabilities. Thanks to such advanced LiDAR systems, Aurora says it was able to reduce required human intervention 18-fold during its test runs in Texas. 

Another interesting application of Aurora’s AI technology is in “robotaxis” by partnering with companies like Uber and Toyota. This past March, Aurora unveiled the first fleet of custom-built Toyota Siennas that can operate at road speeds of up to 70 MPH and are currently being tested in the Dallas area. After the testing is complete, the company hopes to launch a ride-hailing product called Aurora Connect in 2024. 

Einride 

Einride is currently working on a track called Pod but is also notable for its smart electric trailer that can be hitched to an electric truck to provide 400 miles of added range along with data on its loads and usage. This year they got permission from the National Highway Transportation Safety Administration (NHTSA) to start testing their autonomous trucks in the US. 

The public road pilot will begin in Q3 of this year at a GE Appliances manufacturing facility, building upon current existing operations in place with Einride. The autonomous Pod trucks will operate on public roads complete with mixed US traffic to allow for the testing of real-life workflow execution. This will include day-to-day operations such as the movement of goods and coordination with teams at multiple warehouses to ensure efficient loading and unloading.

Embark Trucks 

Embark is focused on the software that makes trucks drive themselves as well as on establishing a network of rock-solid routes on which they’ll do so. Partnerships with Volvo, International, Freightliner, and Peterbilt suggest a wide net being cast to deliver Embark’s promised 10% fuel savings, 40% time savings, and 300% revenue growth per truck. Beginning in February, the company’s testing in Montana utilized Embark-powered trucks traveling on a 60-mile round trip route on public roads between Clinton and Missoula, Montana, in varying winter weather situations. In addition to on-road testing, Embark developed a comprehensive weather model using over 8 billion historical weather data points – dating back over ten years on all major US routes – to analyze the impact of snow at a lane level across the US.

What Types of Data Annotation Are Required to Create Autonomous Trucks 

Autonomous trucks rely on LiDAR technology to recognize other vehicles, road markings, street signs, and anything else they might encounter on the road. LiDAR is a box that is located on top of the vehicle, and it sends out pulses of light that bounce off objects and return back to the LiDAR. The longer it takes the light to return, the further away the object is located. This LiDAR technology produces a 3D Point Cloud, which is a digital representation of the way the truck sees the physical world. 

When data annotators receive a 3D Point Cloud image like this, they would need to annotate the road markings with lines and splines and tag all of the cars and other objects in the image. Additional semantic segmentation may be needed, which involves classifying individual points of a 3D point cloud into pre-specified categories. This task type is used when you want workers to create a point-level semantic segmentation mask for 3D point clouds. For example, if you specify the classes car, pedestrian, and bike, workers select one class at a time and color all of the points that this class applies to the same color in the point cloud. 3D Point Cloud annotation can be a complex and time-consuming task. Trust your data annotation needs to professional companies such as Mindy Support.