What would most revolutionize logistics over the next few years? The most obvious answer is autonomous trucking. But when will this technology become operational? Will it take decades? A decade? Or can we get there in the next few years? That is the core question.
TuSimple – a provider of autonomous freight semi-truck solutions – has said their technology will be operational by 2024. Start-ups have been known to make extravagant claims. But TuSimple has gone public, is now being covered by the equity research firms, and over 5,700 vehicles have been reserved by sophisticated shippers or carriers in just the first four months. TuSimple’s timeline is looking more and more realistic.
TuSimple (NASDAQ: TSP) offers a Level 4 autonomous solution. Level 5 is the top of the SAE autonomy scale; a level 5 solution can perform all driving tasks, under all conditions that a human driver could perform. A level 4 solution can conduct the driving task and monitor the driving environment, but the automated system can operate only under certain conditions.
In TuSimple’s case, “the certain conditions” it will operate in is the “middle mile.” The middle mile is not the “first mile” – the distance traveled between an origin, like a warehouse, and the highway. And it is not the “last mile” – the distance between an exit off a highway to a destination site – a factory, warehouse, or store for example. The TuSimple middle mile will be a terminal near an interstate to a terminal located not too far off an interstate exit.
Overcoming the technological challenges required to operate an autonomous truck has been daunting. TuSimple responded to the challenge be developing 240 patents and testing their trucks over 3.7 million miles. Their technology includes a 1,000-meter perception range, 35 second planning horizon, high-definition maps with accuracy within five centimeters, and a fully redundant sensor suite and components. The trucks are monitored in a cloud-based operations oversight center.
Operating in the middle mile simplifies the technological challenges. It is simpler to operate an autonomous truck on an interstate than the first and last miles. Navigation is also simpler because the trucks will move on defined routes that have been pre-mapped in a manner that the truck computer understands. Further, the middle mile offers fewer “edge cases.”
Edge cases represent situations that could adversely affect driving performance that are difficult to anticipate. In theory, when an autonomous semi-truck encounters a previously unsolved edge case, it will pull over to the side of the road. Or “edge cases” may represent situations that can be anticipated but that are difficult to build math that can effectively respond to the situation. Bad weather is an edge case. Currently, TuSimple is being tested across Arizona, New Mexico, and Texas; all states where snow and heavy fog – difficult edge cases – are less common.
TuSimple is offering freight as a service to shippers and carriers. A customer would purchase TuSimple’s fully autonomous semi-trucks directly from a semi-truck manufacturer. The customer would then pay $0.35 a mile for access to the right to move freight on the company’s autonomous freight network. This freight capacity would come at a 10%-15% discount to the prevailing cost to move freight. Drivers account for 40 percent of the cost of the freight move. The discount is made possible by eliminating driver labor in the middle mile. Based on the driver shortage, the savings being projected may be conservative. Further, autonomous trucks do not face hours of service regulations, the requirement that drivers get a certain amount of rest. Because the trucks will not speed, there will also be fuel savings and a lower carbon footprint.
Safety and regulatory concerns are the main issues that could delay the deployment of the TuSimple technology. While TuSimple has simplified the autonomous driving problem, they have almost certainly not discovered all possible edge cases and built out the math to respond to those anomalous driving situations. In the Risk Section of their initial public filing, the document states “although we believe that our algorithms, data analysis and processing, and artificial intelligence technology are promising, we cannot assure you that our technology will achieve the necessary reliability for Level 4 autonomy at commercial scale… we are still improving our technology in terms of handling non-compliant driving behavior by other cars on the road and low reflectivity objects and performing in extreme weather conditions… There can be no assurance that our data analytics and artificial intelligence could predict every single potential issue that may arise during the operation of our … autonomous semi-trucks.” Safety could also be affected through third party cyber-attacks on their network.
Nevertheless, TuSimple has convinced sophisticated partners, customers, investors, and financial analysts that mature autonomous solutions could be markedly safer than human driven vehicles. According to the National Highway Traffic Safety Administration, 94% of all accidents involve driver-related factors. Carriers are very familiar with all the driver behaviors that lead to accidents – drinking, drugs, fatigue, distraction, driver health issues, aggressive driving, and outright stupidity. An autonomous system does need to effectively perceive the driving environment. But the human health, emotional, and behavioral contributors to accidents are eliminated.
Currently, 24 states allow level 4 autonomous semi-truck commercial deployment. If no tragic accidents occur that result in negative publicity, and that this in turn does not cause the regulatory landscape to tighten, the year 2024 will be the start of the autonomous driving race to replace human drivers. But even in 2024, these trucks will not be running across all lanes nationwide. Rather there will be a focus on delivering across targeted lanes for select customers. The autonomous trucking revolution is going to happen. And it is going to happen sooner rather than later.