One of the cool things about being a consultant to the automotive industry is that I get exposed to a lot of different parts of this business. Recently, I gave a talk on vehicle automation at a Finished Vehicle Logistics conference, which was eye opening for me. Specifically, there are unique opportunities for vehicle automation in the vehicle logistics part of the industry.
Finished Vehicle Logistics is that part of the business responsible for getting cars from the end of the production line to the dealer’s show room. When working for an OEM, my whole focus was on getting vehicles to the end of the production line, ready to go to market. How they got from there to the showroom was a bit of magic that I never thought much about. There is a whole industry moving cars by trains, ships and trucks all across the world. This is generally paid for by the destination charges added on to your bill when you buy a car (at least in the US), typically $400-$1000. This money goes straight to the OEM, so they might want to save a bit here. One of the things this money pays for is a lot of labor, some of it expensive union labor, moving the car between various conveyances and storage locations. Those storage locations can be expensive, and packing vehicles tightly can save a significant amount of space. A lot of space can be saved if the doors don’t need to be opened, and every time a car is touched, there is a risk of scratching. However, someone may need to retrieve the, say, yellow car in the middle of the lot on short notice, which is a major pain if 500 cars need to be moved. An additional function these guys provide at trans-shipment points is the inventory buffer between the highly predictable production lines and fluctuating dealer sales.
Consider one of these trans-shipment points. There are thousands to tens of thousands of brand new cars in a large lot. Possibly from a single OEM, but more commonly from a mix. A ship docks with up to 8,500 cars on board, which roll off to parking spots in a steady stream driven by longshoremen being shuttled onto the ship. After some sorting and possibly a few moves, cars are marshalled into another stream flowing onto another ship, a train, (holding as many as 2,000 vehicles), or a car hauler (truck). There is a steady flow of drivers being ferried the other way, a lot of them trying to find a specific vehicle among the nearly identical thousands. The yard is controlled, no cows or student drivers, and infrastructure can be built and maintained as required. This seems like the ideal place to test vehicle automation and short-range connectivity in a high density, low risk environment.
OEMs build cars for the end consumers and are not going to put a lot of cost in cars simply to automate the logistics around delivery, however, if the hardware is already there and the OEM can save a few bucks with software, which is mostly already developed, this might motivate them. Here are some easy wins.
Here I am! V2X communications (see my earlier post) broadcasts a message with an anonymous ID over a range of several hundred meters which includes the vehicle location. It is technically easy to enable this technology to broadcast the VIN (a unique vehicle ID) to determine the location of every car in response to a query. This would need to be disabled prior to sale for privacy reasons. If the vehicle also reports battery state, and even fitment and destination, this could help find a car among thousands. V2X is ideal for this since you get all the vehicles within range, unlike cellular techniques where some geo-lookup is required. This increased power drain would increase the need to manage the battery, but a remote start command could help with this.
Follow me! A lead vehicle could be driven manually with others following automatically close behind. This is an extension of the existing traffic jam assist app, where essentially a car follows a lead vehicle, and could be implemented fairly simply. More difficult would be managing the queuing- e.g. who follows whom, and marshalling the vehicles into a queue and managing the following lengths, etc. All doable using communications. Applications like this have been demonstrated in high speed platoons.
Self parking. After playing follow the leader, the individual vehicles can park themselves according to some instructions (again using communications). Either just next to the forward vehicle (left or right side), or right behind for a train type situation or large lot. In addition to saving labor, this has the potential to save a lot of space since the door does not need to open. (How do the drivers get out after parking in a train car anyway?) Many cars can park themselves now; we just need to work out the instruction set telling them where to go and when.
Full automation. Within the confines of the yard, and with the aid of a good map. This sort of driving is the bread and butter of automated vehicle research today, the new feature being a map that covers the inside of a ship or train. There have been some recent demonstrations of this application in garage situations (automated valet parking). Here is a chance to test the interactions among a lot of automated cars of different brands, without putting the public at risk. This could come in really handy when the desired yellow car is in the middle of a sea of tightly parked cars. The cars themselves might be able to work out how to best extricate a particular vehicle.
Public roads. The end of the production line is an onramp to the freeway, with a marshaling yard at the other end of the road. The individual cars take off on their own, exiting the freeway at special offramp suitable for automated cars. The offramp leads to a facility where the car can be refueled, or wait for a human driver to take it the “last mile” on roads not yet ready for automation. This technology is close to here today. We need some social and political engineering, but this is probably not far off.
All of these applications are along the automation path we are travelling, and could provide some good automation experience at high vehicle density and little risk to the public. This would require the definition of a message set to enable these applications, but they could easily be implemented on top of the V2X short range standard being deployed. There would be little or no additional vehicle cost, and some savings on the logistics. Putting all the protocols in place, and enabling, say, a Toyota to work with a GM might be complicated, but the industry needs to figure this out if V2X is ever to achieve its potential. Good practice. The Finished Vehicle Logistics companies should start work on a standard for this, which will be needed if anyone is going to get worthwhile economies of scale.