By Robot Rabbi
Lior Ron has had quite a ride since founding Otto less than a year ago; within three months of his official launch, Uber acquired the autonomous truck startup for more than $680 million. Ron, a former Google executive, told me this week that he started Otto as a socially conscious company to end roadside fatalities. Since trucks are the cause of the most catastrophic accidents, it was a natural progression to begin with converting big rigs. A big technological advantage of highway driving is that there are no pedestrians, bikes, and other obstacles that caused an Uber’s self-driving taxi accident this week. By traveling from exit to exit, Otto became the first company to make an autonomous delivery of 50,000 cans of Budweiser last October.
Since then, there are a handful of other upstarts targeting the eighteen-wheeler market. Last month, Embark became the latest to publicly reveal a self-driving rig with the announcement of their Nevada beta test. At their core, Embark, Otto and others are deep-learning companies that utilize sensor-based data (cameras, radar, GPS, LIDAR, and other inputs) and neural networks to enable vehicles to learn the same way teenagers do – practice.
According to CEO, Alex Rodrigues, “Analyzing terabyte upon terabyte of real-world data, Embark’s DNNs (Deep Neural Nets) have learned how to see through glare, fog and darkness on their own. We’ve programmed them with a set of rules to help safely navigate most situations, how to safely learn from the unexpected, and how to apply that experience to new situations going forward.”
Another benefit of tracker-trailers is the ability to platoon multiple ones together in a wagon train. Last year, the US Army utilized this approach to lead four rigs through seven miles of Michigan highway traffic. The trucks were driven semi-autonomously by utilizing vehicle-to vehicle (V2V) communications between the lead human driver and the four self-driving drones. The Federal Government plans to open up 5.9 Ghz for V2V communications in the near future. To accomplish this successful test, the US Army’s Tank Automotive Research, Development & Engineering Center is working with Peloton with the expressed goal of ferrying goods through war zones.
In 10 to 15 years, Army engineers say fully autonomous truck convoys will be ready to serve in conflict zones. The reasoning is obvious: “we do want to get soldiers out of the convoy vehicles, in case they could be on roads with IEDs,” says Alex Kade, who helps direct the Army center’s research in ground vehicle robotics. Since 2002 in Iraq and Afghanistan thousands of military and civilian personal have been killed in transport runs.
Long-haul truck driving is only one aspect of the logistics pipeline, last mile delivery costs retailers and shippers annually over $70 billion. Currently, every logistics supplier from Amazon Prime to UPS are investigating ways to reduce costs and increase margin in the world of free shipping. UPS became the first company in the US to fly a delivery drone autonomously from truck to customer last month. While the milestone flight was a technological achievement, it did raise questions whether aerial drones are the most efficient means of delivery as it took considerably longer than a human driver.