Though it certainly will be exciting to potentially have GPR added into the mix of sensors used for achieving driverless cars, let’s be cautious in overstating the possibilities.
I do not want to seem downbeat on GPR, and to be clear it is indeed a handy addition to self-driving car capabilities, but it won’t radically change things and will take likely years to reach any viable and widespread adoption.
For those pursuing GPR, please keep going, full steam ahead.
Just know that the imagined gold that could be unearthed via producing and selling GPR for self-driving cars is not so easily grasped and some distance off in the future.
And, for those of you unfamiliar with GPR, don’t be fooled by those that seem to imply that GPR will replace other sensory devices used on driverless cars.
Some of the media have gone so far as to suggest that GPR will replace cameras on driverless cars, which is a fundamental misunderstanding of what GPR does, plus a misunderstanding of how self-driving cars work.
The other facet to keep in mind is that even once GPR is better readied for real-world use in cars, it won’t necessarily be tossed into all driverless cars, and instead potentially used on some brands and models, but not all.
That might seem to undercut to the market size potential, though there are going to be other uses of GPR and not solely designated for the advent of driverless cars (thus, GPR has other markets and opportunities to have strong growth).
Explaining GPR Capabilities
Here’s what GPR does.
Using an electromagnetic sensory device, radar beams are broadcast downward from the surface into the subsurface. Based on reflections from the beams, the GPR tries to ascertain what might be underground, such as rocks, soil, roots, etc.
You might be puzzled as to why a self-driving car would care about the subsurface aspects (it’s not as though an everyday passenger car has a digger or shovel that is seeking to dig up the road).
Essentially, the nature of the subsurface area detected is relatively unique and can be patterned into a signature or fingerprint.
If someone were to drive around and collect GPR data on existing roads, they could collect together the data and have a kind of descriptive signature of the roads, providing a map that’s based on underground data.
Then, later on, any self-driving car that was unsure of what road it was on or where it might be on that roadway as to the location, could do a real-time snapshot of a GPR imprint as the driverless car was proceeding, and then compare the imprint to pre-mapped signatures, allowing it to reasonably conclude that whatever matches to the map is the likely position of the self-driving car.
It seems to some like a lot of work when you can just use a camera to “see” where you are or use a GPS or some other technique and refer to a conventional map.
Yes, the GPR is an “extraordinary” approach and not especially intended to replace those other methods of location identification, but it does come in handy in certain circumstances.
Okay, so GPR offers a predominantly augmented sensory capability and would find uses in somewhat narrow circumstances.
There is a bit of a rub.
If the underground signature or fingerprint hasn’t already been mapped, the GPR won’t really do you much good.
That’s part of the odd twist.
Presumably, you would more than likely be using the GPR on rural roads, back-roads, ones that are bound to have snow on them and for which snowplows are a rarity, but you then need to ponder whether anyone would have gone to the trouble to pre-map those roads so that they can be used for GPR in a driverless car.
A chicken and the egg conundrum.
There is obviously a cost and effort required to go around and do the GPR mappings.
Sure, maybe for some popular roads it would be worth the cost, while for less-traveled roads and ones off the beaten path, it would seem not quite as profitable to make such maps.
Thus, a self-driving car would have to be able to access above-ground maps, which are commonly available, and also underground maps, which don’t exist today on any widespread basis and will need to be created.
Not wanting to crush the dreams of GPR, consider too that rain impacts the soil and cause difficulties in matching a real-time snapshot imprint with pre-recorded subsurface signatures.
Depending too on the type of roadway surface, there can be issues in penetrating downward, and the data collected might be noisy or partial in nature.
On the other hand, there are some interesting other potential uses.
For example, a multi-level parking garage could potentially be mapped for its roadway signatures (based not on soil but instead on the various concrete and steel elements that were used in the construction), and then a GPR used in a self-driving car could more readily navigate within the garage.
Yes, a nifty possibility.
Though, on a cost-benefit mindset, would the existing sensors be sufficient anyway, and would the added cost for the GPR outweigh this and its other potential use cases?
Time will tell.
Overall, the question arises: To what degree will the advent of AI true self-driving cars be likely to add GPR into the mix of their sensor suites?
True self-driving cars are ones that the AI drives the car entirely on its own and there isn’t any human assistance during the driving task.
These driverless vehicles are considered a Level 4 and Level 5, while a car that requires a human driver to co-share the driving effort is usually considered at a Level 2 or Level 3. The cars that co-share the driving task are described as being semi-autonomous, and typically contain a variety of automated add-on’s that are referred to as ADAS (Advanced Driver-Assistance Systems).
There is not yet a true self-driving car at Level 5, which we don’t yet even know if this will be possible to achieve, and nor how long it will take to get there.
Meanwhile, the Level 4 efforts are gradually trying to get some traction by undergoing very narrow and selective public roadway trials, though there is controversy over whether this testing should be allowed per se (we are all life-or-death guinea pigs in an experiment taking place on our highways and byways, some point out).
Since semi-autonomous cars require a human driver, the adoption of those types of cars won’t be markedly different than driving conventional vehicles, so there’s not much new per se to cover about them on this topic (though, as you’ll see in a moment, the points next made are generally applicable).
For semi-autonomous cars, it is important that the public be forewarned about a disturbing aspect that’s been arising lately, namely that in spite of those human drivers that keep posting videos of themselves falling asleep at the wheel of a Level 2 or Level 3 car, we all need to avoid being misled into believing that the driver can take away their attention from the driving task while driving a semi-autonomous car.
You are the responsible party for the driving actions of the vehicle, regardless of how much automation might be tossed into a Level 2 or Level 3.
Self-Driving Cars And GPR
GPR could be used in conventional cars and provide some form of visual display or audio indication to human drivers, serving as a secondary mapping system that offers guidance to drivers.
Perhaps some automakers will decide to include GPR into their Level 2 and Level 3 vehicles.
Besides providing a handy capability, the automakers choosing to use GPR would be able to tout that their cars are differentiated from those on the market that lack GPR.
If that did occur, the odds would seem that the other automakers might follow suit, in which case GPR would gradually become commonplace on most cars and become an assumed and everyday expected capability.
That would be a boon to GPR.
Once again, the added cost though for providing GPR would need to be ascertained and whether passing along those costs to those desirous of buying a car would “overprice” a car and hamper car sales.
In terms of true self-driving cars, since they rely entirely on sensory devices for detecting the driving environment, it would seem to some that it’s a no-brainer to add GPR to any Level 4 and Level 5 driverless car.
The mantra oftentimes seems to be that the more the merrier when it comes to having various sensory types on a self-driving car.
Life isn’t that easy.
The more sensory devices you add, the greater the cost of the vehicle.
Plus, think too about the ongoing maintenance and repairs associated with any sensory devices included in a driverless car.
You also need to consider the added weight to the car, and where you can even find a spot to place the sensory package.
Sensors on a driverless car cannot just be randomly placed into the body or structure of the car. There are heat issues to be dealt with, along with routing electrical and communications cables to the device, etc.
Existing GPR devices are today rather large, bulky, and weighty.
Until the size comes down, and the weight, and the cost, and other such factors improve, it is pretty much a no-brainer to not include GPR, at least until it is readily viable for adoption on a widespread basis.
And, as emphasized, GPR won’t be acting alone on a self-driving car.
This means that you need to integrate the GPR detection into the sensor fusion efforts, and figure out when, where, and how the GPR capability will square with the sensor efforts of the cameras, radar, LIDAR, ultrasonic, and whatever else is already loaded onto the car.
Rather than throwing the kitchen sink of all sensors onto a driverless car, it is likely that gradually the automakers and self-driving tech makers will begin to winnow down the set of sensors that are sufficient and complete to drive the vehicle.
You might say they are seeking a parsimonious set of sensors.
Or, more plainly, the Goldilocks set of sensors, not too many, not too few, and instead just the right amount.
Will GPR make that cut?
GPR is not “new” per se.
Devices to scan what’s underground have been around for years, used for exploring the subsurface on other planets and for earthly purposes of trying to detect landmines.
Cleverly, there has been a realization that GPR tech could be used for aiding driverless cars.
That is sensible and will generally be welcomed.
There are some hitches or hoops that need to be jumped through to get GPR ready for prime time use on self-driving cars and will take time to get ironed out.
One big question is whether the ship might sail before GPR is ready.
In other words, if the automakers and self-driving tech makers have proceeded along without GPR, and meshed together with their existing sensors into a coherent and rolled-out driverless car fleet, will they be willing to essentially retrofit their vehicles to accommodate GPR once it becomes viable?
Look into the future.
If self-driving cars are roaming around successfully, doing so without GPR (since it wasn’t yet available for use when those driverless cars were first turned into production), why would an automaker go the trouble to add GPR?
It seems somewhat unlikely that you’d try to add it to the fleets already in existence, though of course such a retrofit would be possible at some likely exorbitant cost and logistics nightmare possibility.
The odds are that it would get added into the next-gen of self-driving cars instead.
Yet, there is still a cost to do so.
Besides the cost of the GPR device itself, there would be the cost of integrating the GPR into the vehicle and the cost of integrating the GPR into the systems that deal with sensor fusion.