Tow trucks, semi-trucks, and tractor trailers share a common purpose – they exist primary to transport something from one location to another.

They also happen to be just about everywhere.  With close to 2 million semi trucks, 6 million semi trailers, and some 32,000 local towing company drivers in the US alone, they are a formidable presence on our highways and roads.

The trucks and the drivers that operate them are a critical component to keeping our goods flowing, our roadsides clear of accident, and are foundational for a high functioning economy. While driving a truck, or towing a car is not something that the author has have done, we as AI developers in the driving space would be remiss in leaving this important segment out of consideration when developing a robust AI strategy for a driver-less future.

For a seasoned towing service or long haul driver, towing or transporting a load is a relatively simple series of steps, however using the lens of AI logic, it is enormously more complex than the already ridiculously complex logic of driving a car safely without a driver.

In this week’s post well talk about the complex logic behind towing a load, how it is different than standard driving, and what would need to happen to have a future with automated tow trucks and logistics.

Why Towing Is Complex AI

In the world of IT an “edge problem” is an atypical use case, which should be addressed, but considered a lower priority than the primary consideration. Towing itself is an edge use case that itself creates a number of new edge problems for the AI, as it has to consider both the vehicle and the trailer or load it is towing. The AI not only has to be able to operate the driving vehicle, but the logic also must account for the status of both vehicle and load at the same time, all the while adjusting for the changing dynamics of each component.

One of the biggest problems with towing is the fact that the car and the trailer or truck are now two separate vehicles. The operating vehicle must be able to control both vehicles, and it also has to be able to navigate around obstacles. If the car brakes suddenly, the trailer or truck could jackknife and cause a serious accident.

Another problem with towing is the fact that the car and the trailer or truck have different centers of gravity. This can cause the car to sway back and forth, which can be dangerous on the highway. The AI has to be able to compensate for this difference in center of gravity, otherwise the car and load could tip over.

The AI also should be able to navigate around obstacles. Towing a trailer or truck creates a number of new obstacles, because now the car must avoid both the trailer or truck and the obstacles in front of it. If there are cars parked on the side of the road, the AI must be able to maneuver around them without hitting them. If there are construction barriers, the AI should be able to go around them.

In addition, towing creates new wind resistance and increases drag on the car. This can affect the car’s handling and braking abilities. Towing also increases fuel consumption because the engine has to work harder to move both vehicles.

The Logic for Towing A Load

A typical towing service call requires much more simply hooking up the car and driving away. Towing a load requires a complex algorithm to ensure the safety of other vehicles, pedestrians, and of the load being towed.

While there are thousands of small considerations the tow truck driver makes while driving, here’s a few that would need to be automated in order to begin to duplicate a towing scenario.

Before the trip even begins, the self-driving car or truck must first determine if it is safe to do so. This means considering the weight of the load, the size of the vehicle, and the condition of the road. If it is determined that towing is unsafe, then the AI will need to find an alternate solution or cancel the load.

Once it has been determined that towing is safe, the car or truck will need to calculate the best way to tow the load. This includes determining the speed at which to travel, the braking distance, and the turning radius.

Braking distances and turning radius would need to be accurately assessed and adjusted according based on the weight and dimensions of the load.

While on the road, the truck will need to continually assess the dynamics of the load and adjust or cancel the drive safely if necessary. Wind resistance can significantly affect the drive of a truck and trailer, especially with perpendicular, high velocity winds when the loads are light and large. Truck drivers will instinctively drive slightly into the wind to adjust for the force pushing the load on way or the other, the driving AI logic must be able to do the same.

Once at the destination, the load will need to be dispatched. This includes determining the best way to unload the cargo and managing traffic around the area.

All of these considerations are what is known as “edge problems” or “pocket cases.”

Adjusting Vehicle Communication

Clear, effective communication is a important aspect of safe driving. In a standard driving situation, a truck driver is able to communicate with their turn and brake signals, but also via hand signals, eye contact, and verbal cues.

This communication changes in a towing situation, say for example – a diesel truck towing a fully loaded 53′ trailer. A vehicle and load this large require that the truck driver will need to indicate their need to turn much further ahead that of a smaller vehicle, it also requires much of a nuanced approach when merging on the highway.

With this loss of human cues, communication needs to much more robust than the current brake and turn signal schema. A potential solution will need to involve a combination of full vehicle sensors that connect the trailer or load to the operating vehicle, clearer turning signals, and other bright lighted indicators involving a color-code for actions being taken by the truck.

Complex But Solvable

As we have seen, towing a load may seem like a simple task, but it is anything but.  While we are getting closer and closer every year to an automated future the complexities of edge cases involved with businesses like a towing service and a logistics or transportation company will need to be addressed for us to enter a confident future where driver-less vehicles are common and accepted as safe.