Otto Set to Disrupt the Freight Industry
Otto's self-driving technology will disrupt the $1.4 trillion freight industry
The American transportation industry accounted for 8.6%, or $1.4 trillion of annual US GDP in 2013, making it the 11th largest industry by GDP. [1][2] The trucking industry accounts for 81% of the transportation industry. Despite its size, the trucking industry has seen little innovation over the past century surrounding its business model. Freight movement is still driven by trucking companies, dispatchers and drivers. Specifically, companies that wish to ship goods access a network of trucks via a dispatching service (either 3rd party or wholly-owned), which matches capacity with demand. Once the match occurs, drivers physically move the products from point A to B. Drivers are hampered by hour limits, meaning that trucks cannot be fully utilized given driver protection laws. [3]
While there has been some innovation along the edges (better data collection via platforms like Lytx to improve driver safety and reduce insurance costs), none of these innovations address the core operating model: drivers, trucks and dispatchers. [4]
Enter Otto – a self-driving truck company. Otto retrofits trucks with hardware kits that enable trucks to navigate, steer and accelerate without human intervention. Otto accomplishes this through both hardware and artificial intelligence. Its hardware includes LIDAR (light detection and ranging) technology, which surveys its distance to objects by sending and receiving laser signals.
Otto has built sophisticated artificial intelligence to interpret these signals to make decisions on what to do: brake, steer, accelerate, etc. Human-based decision-making is very complex, especially during driving. Humans react to new situations using context they have gained via previous experiences. Artificial intelligence mimics human decision-making. Using large data sets, artificial intelligence algorithms build up databases of decisions that it can draw upon during new experiences. These new decisions in turn strengthen the underlying algorithm. In other words, these algorithms are not deterministic – humans cannot code algorithms to cover the millions of situations trucks will encounter. Rather, the algorithms are learning – they make decisions and use those decisions in turn to improve the algorithm.
Otto’s technology, then, aims to disrupt the $1.4 trillion dollar industry by removing the bottleneck: humans. Without driving limits, truck utilization can essentially be doubled by unlocking more hours trucks can spend on the road. Not only will this speed up the time needed to transport goods from point A to B, but allows more goods to be transported over a comparable time period. In the best-case scenario, trucks can then transport the same daily average of 55 million tons using half the truck volume.
Otto is also betting that it can bring this technology to market faster than self-driving cars by being slightly less-ambitious. Otto’s first generation of driverless technology will only work on highways (e.g., “between offramps”). [5] This cuts out many of the decisions that the algorithm will need to make, such as stoplights, crosswalks, stop signs, pedestrians and stop-and-go traffic.
Lastly, Otto’s self-driving technology represents a significant environmental benefit. Truck emissions account for 12.5% of the United States’ annual output on an annual basis. [6] A typical truck has a mileage rating of 6-8 miles per gallon (hybrids, on the other hand, have a mileage per gallon of 56). [7] Decreasing the variability in acceleration will reduce emissions by at least 10%. [8]
To be explicit about their business and operating models: Otto doesn’t yet have a business model, as they’re focused on building the underlying technology. There are many methods to choose from to monetize the technology, e.g. a flat fee for the kit, a usage rate based on number of miles driven, etc. Otto’s operating model has two aspects: delivery and improvement. On the delivery side, their kits are modular, making them very easy to install in existing trucks (as opposed to building trucks from scratch). Second, once in the trucks, they can consume data to improve the underlying algorithms.
While this represents a significant improvement over the existing trucking industry model, they can further innovate by building a capacity marketplace. In addition to their kits, they can ensure high utilization of space within truck trailers by matching capacity to demand within routes – much like uber matches supply and demand in UberPool. This represents an opportunity to disrupt the dispatcher-side of the business in addition to the driver side while also further maximizing capacity and utilization. (717 words)
[1] “Freight Facts and Figures 2015”. Bureau of Transportation Statistics, http://www.rita.dot.gov/bts/sites/rita.dot.gov.bts/files/data_and_statistics/by_subject/freight/freight_facts_2015, accessed 11/18/2016
[2] “Reports, Trends and Statistics”. American Trucking Association, http://www.trucking.org/News_and_Information_Reports.aspx, accessed 11/18/2016
[3] “Summary of Hours of Service Regulation”. Federal Motor Carrier Safety Administration, https://cms.fmcsa.dot.gov/regulations/hours-service/summary-hours-service-regulations, accessed 11/18/2016
[4] “Lytx DriveCam™ Helps Andrus Transportation Services Follow Its Core Value of ‘Doing the Right Thing’” press release, 11/7/2016, PR Newswire, http://www.prnewswire.com, accessed 11/18/2016
[5] Steinmetz, Katy. “Inside Otto, Uber’s New Self-Driving Truck Division”, Time Magazine, 8/18/2016. http://time.com/4458507/otto-uber-deal-driverless-autonomous-trucks/, accessed 11/18/2016
[6] Valentine, Katie. “Big Trucks Emit Huge Amounts Of Carbon Every Year. The EPA Is About To Do Something About It.” Think Progress, 6/2/2015, https://thinkprogress.org/big-trucks-emit-huge-amounts-of-carbon-every-year-the-epa-is-about-to-do-something-about-it-5c402732888#.kj7knkl7n, accessed 11/18/2016
[7] Roberts, Jack. “The Future of Fuel Economy”, truckinginfo.com, 6/2016. http://www.truckinginfo.com/article/story/2016/06/the-future-of-fuel-economy.aspx, accessed 11/18/2016
[8] Enstein, Paula. “Autonomous and Self-Driving Trucks Will Improve Safety, Fuel Economy”, Trucks.com, 4/7/2016, https://www.trucks.com/2016/04/07/autonomous-trucks-improve-safety-fuel-economy/, accessed 11/18/2016
Shezaad – the disruptive power of entirely self-driving trucks is apparent, but how will the business model work given the software is only designed for highway use? Doesn’t this mean that a paid driver will still need to sit in the front seat to monitor the truck while it’s on autopilot, and to take over when it gets to the off-ramp? Is the bet here that these drivers will not be subject to the same hours regulations as those of fully manual trucks, given their dramatically decreased workload? Otherwise, it seems to me that the only thing Otto achieves is potentially safer and more gas-efficient highway driving, but without really cutting out the real bottleneck – the driver.
On a totally separate note, it sounds to me that Otto’s AI set-up is very similar to what Google is doing as part of their self-driving car initiative. Isn’t it inefficient for multiple companies to come up with their own machine-learning AI systems based on their own data, when just one system leveraging all existing data would arguably be best? Do you think this market could go open-source in due time? Finally, I wonder if by acquiring Otto Uber actually seeks to get into trucking, or if it will just re-purpose the software to enable their own self-driving cars (or both!).
While I agree that there are many benefits of self-driving trucks, including improved road safety, one concern I have is the impact this transition will have on current employees. An NPR article last year indicated that truck driving is the most common profession in a majority of states (http://www.npr.org/sections/money/2015/02/05/382664837/map-the-most-common-job-in-every-state). Eliminating this profession will exacerbate the economic pressures on low to middle income workers. As evident with the presidential election, many workers are already feeling under threat as their jobs are being eliminated. In parallel with developing this technology, companies should be considering how they are going to transition their employees to new jobs either inside or outside the company. By developing necessary skills now, they can smooth the impact self-driving trucks will have on the broader economy.
This is an amazing piece of technology and I’m really excited about the work that Otto is doing in this area. The trucking industry is facing a shortage of drivers and if current trends continue, the shortage is expected to rise to 174,500 drivers by 2024.[1] With a shortage of drivers, transportation costs have steadily increased and the ability to do just-in-time and last minute shipments has decreased. Truck carriers hold significant leverage over companies’ supply chains and are able to more strongly dictate terms of service and payment. Large companies are able to weather this by a combination of passing costs on and adjusting their supply chain but smaller companies will inevitably suffer.
I think what companies like Otto should do is not only pitch this to the traditional truck carrier company, but also pitch this to companies with large supply chains (and maybe their own truck fleet). By making this part of the supply chain autonomous, companies that traditionally may have outsourced this operation to truck carriers, would possibly now be interested in investing in the capital to own and more directly control this aspect of their supply chain.
[1] http://www.trucking.org/ATA%20Docs/News%20and%20Information/Reports%20Trends%20and%20Statistics/10%206%2015%20ATAs%20Driver%20Shortage%20Report%202015.pdf
Thanks for the post Shezzad! As a new driver I have to agree that highway driving is far easier than local driving, and given the clear economic advantages you mentioned (higher truck utilization, lower driver cost), I tend to believe that self-driving innovations will hit the trucking market much harder than the consumer market once it gets there.
I did have two questions for you on the Otto model–how it deals with regulatory issues and how it expects to train its AI. In contrast with cars, trucks have incredible destructive potential and variability in weight, size and purpose. When Tesla’s AI fails, one driver is killed–however when a truck’s AI fails, it could potentially kill dozens of drivers. How has Otto addressed these safety concerns as they attempt to roll out the self-driving platform? I believe self-driving trucks face a much higher regulatory hurdle than self-driving cars will due to this.
Second, it seems like the AI algorithm needs to be trained to learn how to deal with situations. How does Otto plan to do this training? I know Google basically had private roads on which its self-driving vehicles were trained, but curious if the need for highway-type scale, speed and variances make this approach unwieldy.
Shezzad – great post! The freight industry is clearly a very large industry that is prone to disruption from self-driving vehicles, but I still wonder if Otto is going to be the one to do it. I get that they are focusing on a very niche part of the self-driving technology, i.e. driving on highways. However, it seems like the more advanced self-driving technology that others are developing (Google, Tesla, GM, Uber, etc) can be easily adapted to service this market as well.
Thank you for opening my eyes to the world of driver-less trucking! While I am skeptical about the execution of an actual self driving truck that can account for the infinite number of nuances an data points that need to be considered when making a split second driving decision, I am hopeful that it can be done because the inefficiencies in the freight trucking industry are abound. This is an industry that has built its norms off of “that the way we have always one it” and it will into change without the help of a disruption grenade like this one. What I would like to know are what percentage of the roads or major cross-country shipping routes in the US are predictable enough where you could confidently eliminate a tenured experienced driver. One way or the other, this technology will happen, now its a matte rof who gets their best and first.