There are a lot of “bits and pieces” in this segment. Last year saw a decent uptick in both technology and test beds of level three (mostly) and four cars. For those keeping score, drivers are still necessary in Level 3 the vehicle will prompt the drive to take over in certain circumstances, while Level 4 is fully autonomous under select conditions, according to Car and Driver.
However, the low hanging fruit has been picked. We are currently stuck around level four out of six (or five, depending upon who is defining). The next few years will focus more on overall expanding the current state-of-the-art and perfecting what is already in the pipeline. The final levels will require external interfacing.
Therefore, autonomous vehicles, in 2018, will be much like the Internet of Everything/Everyone (IoX) – MOTS, and we will see more hype than substance. The next level of real progress will occur when two-way communications are integrated. That involves a whole lot more than just equipping vehicles with more and better sensors, and improving the capabilities, which is the present state-of-the-art.
To reach the lofty goal of the final level of fully hands-off vehicles will demand five-nines of integrity. I will wait for the 99-nines, myself. That means full integration with the infrastructure and environment in terms of reliability, functionality and ubiquitous interconnect. One-way sensors have come a long way with their pervasive implementation in smart cars, but sensors alone are just not sufficient.
Several subject matter experts (SMEs), including one at California Path that I have known for many years, make the case that before we reach full autonomy, connected cars must have bi-directional communications with everything around them. That includes each other, as well as infrastructures, other moving objects, and the cloud (web) to enable advanced metrics in safety and efficiency (acronyms are; vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and vehicle-to-everything (V2X). There are also additional elements that are not commonly mentioned, such as vehicle to pedestrian (V2P) and vehicle to bicycles and scooters (V2BS).
Rhode and Schwartz presents similar data in a recent report that talks about Advanced Driver Assistance Systems (ADAS).
“Advanced Driver Assistance Systems are clearly making a difference in vehicle safety as more cars include them and as drivers learn to use them.” However, note that the next sentence is: “Further gains in safety are possible by adding a communications capability to every vehicle and the surrounding infrastructure.” Here is the bi-directional issue again, and that will take many more years.
Another sector, trucking, is also working towards automation. In a recent interview given by Derek Leathers, president and CEO of Werner Enterprises, he said that it will be at least 15 to 20 years, if ever, before a fully autonomous truck will be rolling down the highways.
Finally, as Jada Tapley, Delphi’s vice president of advanced engineering, notes “if you don’t have the right architecture in place to enable those sensors and support all the additional software and computing power that’s necessary for a level four or five application, then it’s just fundamentally not going to work.”
I mentioned earlier that there will be a lot of hype. That is a main spoiler because there just is not the technology to do what many, from auto manufactures to elite players (Tesla) to software and hardware players, are laying out as existing. Reading between the lines, I always find exceptions or exclusions. I say that because I hear so much about the platform, but when I dig deeper, there is relatively little tangible development going on at these next levels.
Then there is the uber-hype from companies like Nvidia. They claim that they have a processor (named Xavier) that can achieve full level five autonomous driving right now. They are using terms such as “AI car supercomputer” and are targeting “robotaxis” as the initial application. Most experts see their claims as unlikely to be deliverable for years, if not decades. These types of ridiculous claims do not help the industry maintain credibility.
Other spoilers include things like liability. On that issue, malfunctions in autonomous vehicles are not all that variable. If a tower falls over and part of a network goes down (assuming it did not fall on anyone) that is only an inconvenience (perhaps a large one, but still only an inconvenience). With autonomous vehicles, malfunctions are highly likely to result in injury or death. There is a lot of discussion around who will, where and how liability will be handled.
Then there is the case of standards and mandates. Everyone thought that Dedicated Short Range Communications (DSRC) equipment, which was mandated to be in all vehicles by 2023, was a closed case. However, all of a sudden, that mandate was cancelled and there are now players bringing alternative options to the game.
There is a fair amount of discourse across the industry around this vector. The IEEE 802.11p-based DSRC was being pushed as the standard to work with. But of late, C-V2X, is being pushed because it can use existing 4G LTE equipment.