Submitted by Amy Nguyen 10/03/2020
Last March, I had the misfortune of popping two of my car tires on one of Houston’s many potholes. Due to this misfortune, I could not simply get the jack and spare tire out of the trunk and attempt to change the tire myself. This resulted in me having to call roadside assistance. During the initial call I was under the impression that I could have someone come out with an extra spare and I could buy if off of them, but this was not the case. After another phone call to roadside assistance and a two hour wait, a flatbed wrecker tow truck finally came and towed my car safely home. Although the experience was tiresome, it left me to ponder upon an interesting revelation: despite the automation of driving and the development of self-driving cars, there will always be a need for tow truck drivers and drivers within the auto-shipping industry. The process of automating a flatbed tow truck to operate on its own would be too difficult and too expensive to develop.
As I watched the flatbed truck driver load my car onto the bed of the truck, I noticed how complex the operation was. He had to first position his truck appropriately in front of my car, lower the bed and place it at an angle for which maneuvering my car onto his truck would be easy, and hook chains onto my car. This was all prior to moving my car onto the bed of his truck which now sat at an uphill angle! Afterwards through using a sophisticated motor on the back of his truck and driving my car he was able to get my car onto his truck, secure it, and level the bed again. These steps were done flawlessly through his years of experience, and I am not sure that an AI or computer program would have been able to do so. His work required human touch and feel to it.
The number of cameras and sensors needed on that truck would have been astronomical and the AI would have needed to be well programed to move the truck with precision on the busy streets in the area my aunt lived in. Additionally, the truck was quite large and cumbersome and needed to make a number of wide turns and would have to operate under a different set of conditions and rules that other self-driving cars could not. Some of the lane changes the truck made would not have been allowed in other cars, and the AI would also need to be programmed when to optimally take the wide turns the tow truck had to make.
This would have required extra cameras, sensors, programming, and complex AI to gauge the turn. Tow truck drivers need to be able to look and assess over 360º and process information in real time, which would require great computing power from an AI. With that said, it would have been difficult for a self-operated tow truck to be able to load my car onto itself. Another consideration would be that of public safety- the self-operating truck would have to properly secure the one-and-a-half-ton load that is another car onto itself. Improperly securing the car to itself could result in the car becoming loose and falling into traffic which could result in public injury and financial loss to parties involved. If we were to entertain the idea that somehow the tow truck could have safely loaded my car onto itself, the next question to ask would be- would it be able to get my car off of itself safely and placed in a specific spot? The probability of that seems unlikely.
I believe that AI has a long way to go in determining where people want things strategically and safely placed. Something as finicky as a car in a neighborhood with meticulous homeowner association rules would require more controlled, finetuned actions than that of current automated technologies. The technology, research and development, and time required to make a self-operating tow truck would never be profitable nor would it be feasible within the near future. Many companies have already poured billions into the research and development into the production of simple self-driving cars; however, complicating that with an automated car loading and unloading feature alone would require several more billions of dollars just in research and development.
The sensors and cameras that will need to be outfitted on the car will also be an additional several thousand. This case was also for a short distance tow. The case for automated long-distance vehicle shipping presents within itself additional complications. While I understand that there is the potential for automation in some areas of longdistance vehicle shipping, I will be focusing on why the delivery aspect of long-distance vehicle shipping cannot be automated. After my friend had gotten an internship in Cincinnati, Ohio he realized he had to get his car shipped to there from Austin, Texas. He arranged for the service and dropped off his car with a vehicle shipping company.
Due to a clerical error, they dropped off his car much earlier than expected without my friend present. Although a person made the mistake, I doubt an AI would have been able to perform the action of dropping a car off by itself without a person there. There would be numerous factors to consider. Would the car be able to know where to precisely drop the patron’s car off? Would it know where it could legally park the car, or could it distinguish between illegal and legal parking? I’m not sure it would be able to, and if it could, the same challenges presented in earlier parts of this essay would need to be addressed as well. I will not say that the automated shipping and delivery of vehicles is impossible; however, I will argue that the research and development and time required to create a system that can do so will not be feasibly reached within our lifetimes due to the cost and complexity of such a project. As such, there will be a need and demand for the vehicle shipping industry within the foreseeable future.
Submitted by Amy Nguyen 10/03/2020