The yard dog, the diminutive cousin of the Class 8 tractor, is getting an autonomous AI-powered upgrade courtesy of Outrider.
Outrider, an autonomous yard operations leader, recently announced an industry-first deployment of advanced reinforcement learning (RL) techniques across its customer sites. FreightWaves spoke with Andrew Smith, founder and CEO of Outrider about what the release means for the startup and how it’s taking the humble yard dog to new heights.
For those outside the supply chain, yard operations can consist of drayage operations at ports, intermodal operations at railheads, or in the yards from distribution centers to trucking terminals. The humble yard dog is used to shuttle empty and loaded trailers to and from docks to parking places, where the larger Class 8 tractor hauls the goods for longer distances.
Outrider noted in the release that incorporating the RL techniques increased path planning speed by 10x and enabled the Outrider System the ability to move freight more efficiently and safely through crowded, chaotic and busy distribution yards. Smith adds that right now their network represents about 20% of all yard trucks operating in North America with many customers invented in pilot operations since the early days of the company back in 2017.
How the reinforcement learning works
RL is part of the successful application of AI in the physical world, where the AI collects a large dataset that is then used to train the AI. For the yard dog, Smith notes the 5 years’ worth of data collected from their autonomous yard trucks is the primary source to train the system. Outrider boasts over 200,000 safety scenarios while third-party safety experts and its Fortune 500 customers validate the safety cases.
With that training data, the next step is to reinforce positive behaviors and interactions. Smith notes that some preferred behaviors like following traffic rules and maintaining appropriate distances from both a safety standpoint and driver comfort standpoint are two examples where RL can take the software from a traditional 90% there to the mid 90%, with the systems allowing the AI to handle more complex scenarios.
A great metaphor is in manufacturing where you can spend half your resources to get to 80% production efficiency, and then the next half of your resources to go from 80% to mid-90%. In the world of AI and reinforcement learning, the RL part is where you get from 80% to the mid-90%.
“Where we are learning and where we are applying these AI techniques is across the entire set of aspects that are not just driving the truck but all the other manual tasks that have to take place in a yard,” said Smith.
But yard operations do not just involve hauling a trailer across a yard, there’s also the backing, docking, and hooking then unhooking from trailers.
A day in the life of an autonomous yard truck
At a busy hundred dock facility, yard dogs will handle hundreds of trailers a day, as warehouses load, unload, and reposition trailers from warehouse docks to trailer drop yards.
Smith gives one example of how Outrider has automated not just the yard dog, but the warehouse, yard management, and transportation systems to dispatch the trucks. It first begins with an electronic message like, “Hey we need trailer 227 to be put at loading dock 14.”
“What our system can do is autonomously dispatch an autonomous vehicle. It can move around the yard. It can find the trailer using deep learning and confirming the location of every asset in the yard,” said Smith.
The next step is where the magic happens, with Outrider equipping their autonomous yard dogs with an advanced claw capable of robotic manipulation to connect airlines, one of the biggest hassles if you don’t have a driver.
Smith describes the process saying, “It can autonomously hitch to the trailer. It uses a deep-powered robotic manipulator to connect the airlines and electric lines if needed on the trailer. It can then move that trailer through traffic pausing for pedestrians, golf carts, maintenance workers, etc. and then place that trailer with centimeter accuracy and loading dock only after confirming with a proprietary safety system that dock door is closed and the door is safe to be backed into.”
The next step involves unhitching the trailer and disconnecting the airline and brake lines. Smith adds, “so all that movement can now be fully automated with the outrider system and the reinforcement learning we’ve integrated into it.”
The same RL techniques that teach the autonomous yard dog how to navigate the yard, back into spaces and shuttle trailers also extend down to operating the robotic claw which needs tactile precision to hook and unhook trailer air lines.
Smith highlights that while many companies have tried to modify the trailer, Outrider’s system works with modified and unmodified units courtesy of the AI and robotic arm.
“If you really want to go track down every trailer in your yard and modify it, we can connect to modified trailers as well. However, what we can do is with the existing fleet of unmodified trailers constantly flowing through all of our customer’s locations, we can use AI and robotic manipulation to connect airlines, which is one of the biggest hassles of a driver,” added Smith.
Autonomous yard dogs go electric
Outrider’s yard trucks can also go electric. Smith said, “Not only are we doing the fully integrated system with a custom design set of safety mechanisms but we also layer our systems on top of an electric vehicle platform.”
Smith notes that they’re able to help speed up the adoption of electric vehicles with autonomous trucks having the ability to charge when there’s downtime in the yard and knowing when to charge during lower periods of power usage. A common challenge for electric vehicle charging is if you charge during peak transmission time, there’s the potential to be charged extra compared to charging during periods of less activity.
Additionally, the smaller yard truck does not require the range requirements compared to larger Class 8 trucks and day cabs, where mileage per charge is a limiting factor.
Looking ahead 2025 looks to be a big year for Outrider, after securing multiple patent grants and raising a $62 million Series D financing back in October 2024. That brings the total equity capital raised to date to over $250 million.
