Let’s be honest; everyone wants Tesla or Rivian or Google’s Waymo or General Motors to work out the kinks and get self-driving cars figured out. But the bottom line is that we’re not there yet. And for Tesla, that delay may have an effect on its valuation.
There are a few problems with completely autonomous cars. For starters, the technology is not quite there. But perhaps more importantly, the necessary data for cars to predict and respond to obstacles on the road is not yet robust enough.
For self-driving cars to become a reality, in the way we dreamed while watching the Jetsons, all tech companies need more data to build smart enough AI. And not Google, Tesla nor Uber has enough driving data to make autonomous cars work the way they need to.
The short story is, that without enough driving data, AI cannot run effectively. Artificial Intelligence runs by being able to collect data, store those scenarios, build algorithms and then use the information effectively.
That’s one of the reasons for those Captchas you often get; recognition software needs enough data to be able to distinguish a dog from a fox, and you from your sister. AI relies on human-driven data, especially when it comes to self-driving cars.
Who will be first to have a full autonomous vehicle?
There are many car companies racing to be the first to have a fully autonomous car. But for now, there is still a bit of an in-between phase.
Take Tesla for example. They may well be the first, but like Uber, Tesla’s Autopilot still needs a human driver to be in the driver’s seat. And importantly, the driver needs to remain alert the whole time.
So what would it take to get Tesla or electric car manufacturers such as Rivian all the way there?
One idea could be taken from Jaguar. This luxury car line has partnered with IOTA to incentivize data by rewarding its divers with cryptocurrency.
Crypto for Data
Data is understandably a contentious topic; who owns it, and what are these companies using our data for are important questions.
One way to solve this issue is to start offering some kind of remuneration for our data, as though its our intellectual property, which is really should be. It is this kind of fair exchange for information that cryptocurrency and blockchain tech are so well poised to fill.
One such initiative, as I mentioned, is from Jaguar and the non-profit IOTA Foundation. The car company and the non-profit have teamed up to reward drivers for sharing their driving data.
The pilot project is conducted by Jaguar’s software engineering center in Shannon, Ireland. The technology has already been integrated into several vehicles, including the Jaguar F-PACE and Range Rover Velar.
Jaguar has equipped its vehicles with the internet of things (IoT) sensors. These sensors collect data that is shared with IOTA’s Tangle’s distributed ledger. The car sensors communicate driving conditions to IOTA and that data can be shared with the appropriate authorities. The aim is to collect driving and road data, as well as to improve driving conditions.
In exchange for this information, IOTA offers tokens, or a form of cryptocurrency, via smart wallet technology. The users that contribute data can then use their cryptocurrency earnings to pay tolls, parking fees and electric charging on the move, or even coffee. The smart wallets can also be topped up using conventional payment methods.
So what are other car companies trying to do in order to get self-driving cars to the next stage?
In recent years, Tesla has released its newest version of Autopilot. Autopilot is a self-driving application that is not fully autonomous, it is a “driver assist” system that needs an alert driver at all times. By now, Elon Musk had hoped that they would have fully autonomous vehicles for individual use.
Currently, you can buy a Tesla vehicle that is expected to be fully autonomous in the future, or you can purchase the Autopilot application which can be fitted into vehicles. As an external addition, the application needs to be ordered by phone, as it does not appear to be selling on their online stores presently.
Tesla’s current application will essentially self-drive, but the driver must remain alert the entire time. That means the vehicle requires the driver to interface with the vehicle periodically, such as a hand on the wheel, or it will emit warning signals.
Still, Autopilot is able to change lanes for you, and allows for minimal interaction with the car, as the sensors look for a safe space, as well as adjusts the speed accordingly.
Autopilot user and Forbes writer, Brad Templeton states:
“Autopilot is, at its core, the combination of adaptive, or “traffic aware” cruise control (TACC) with lane-keeping and lane-changing. The standard car comes with only regular cruise control, along with automatic emergency braking and other typical ADAS features.”
While Elon Musk promised that 2020 would be the year for fully-autonomous cars. That has yet to be seen, and frankly, there is no firm date that can really be foreseen. Moreover, Autopilot does not work well in construction zones, as there are too many unpredictable data points. There have been serious issues in construction zones, such as running into barricades, as the technology is not ready to handle such uncertainties on its own.
Also, and very importantly, Autopilot does not see red lights or stop signs – that is just one reason that the driver must stay alert.
How Autopilot Works
Autopilot is built using cameras, radar, and ultrasonics. The system constantly learns from other cars in the Tesla fleet and continues to improve.
While the car is driving, the camera, radar, ultrasonic sensors, and GPS all work together. They provide real-time feedback from the Tesla fleet. The data is aggregated and used to improve the overall system.
The system needs to learn for itself and from others. That means the Tesla fleet needs to collect maximal data about road and driving conditions.
Self-Driving Cars: The Hype v. The Truth
Currency self-driving cars have has their greatest success in Arizona and in California. One of the problems is that there is almost an incalculable amount of variables when it comes to driving. That means that collecting the necessary data for all of these scenarios is still a process.
For many, autonomous cars and Tesla’s autopilot are worrying. It’s mainly because there is a disconnect between the machine and the human driving the car, making decisions. And that seems to still be the case when it comes to self-driving cars.
On March 18, 2018, in Tempe, Arizona, Elaine Herzberg was the first fatality of self-driving cars. She was walking her bicycle across the street and was struck and killed by an Uber car with a safety driver behind the wheel.
Uber responding by taking its self-driving cars off the road. But the real concern is that the AI used does not have enough information to understand all of the moving and unexpected objects on the road.
It is difficult to plan for the unknown when you are human, and even harder when you are AI. If the vehicle does not recognize the situation, the object, or the animal, then the car will not necessarily respond correctly, as was seen when Herzberg was hit and killed.
Chalk it up to the challenge of collecting this information. The reason that searching and basic algorithms are so effective is because we use our computers and smartphones for everything. That means all of the data banks have loads of information to draw on.
This is not the case for driving. In fact, aside from your music and your GPS, when you are driving you are likely using your phone a lot less – I hope anyway. So, unless your vehicle is equipped with an IoT system like Jaguar’s or you have paid $3000-$5000 for Tesla’s new Autopilot, companies building autonomous cars do not have sufficient data to draw on. Engineers need more testable hours.
Why The Cars Need Your Data
The research is incredibly expensive for these companies, and for the drivers. There are not that many people willing or able to spend the money on a Jaguar or a Tesla product, where low-end vehicles start is in the range of 6 figures.
Perhaps compensating drivers for data concerning driving habits and road conditions with cryptocurrency is a part of a new trend that we will see. Jaguar and IOTA were the first to team up to offer cryptocurrency. But maybe, given that Tesla has missed this year’s target of fully autonomous cars, they will consider following suit.
If you are not a fan of the Tesla ethos or have no stake in seeing self-driving cars become a reality, then, you will not likely be interested in where this article is going. Either way, I argue in favor of why you might want to consider exchanging your driving data for cryptocurrency.
As I have stated, and Elon Musk is woe to admit, neither the technology or the research is ready for fully autonomous cars. However, both of these issues are related to a lack of data and limited funds.
Still, fully autonomous cars, when they work the way we dream that they will, might do the following:
- Eliminate car accidents
- Give greater mobility and accessibility for those unable to drive
- Free up valuable land now used solely as parking lots
- And my personal favorite, it will allow drivers to nap or read while on their commute!
So what’s standing in the way of self-driving cars and Tesla’s Elon Musk’s, dream?
Driving Data: How It Works
In order for Tesla to get its autonomous car on the road, it needs a data bank to draw on. In that arena, Waymo, Google’s sister company, is unsurprisingly well ahead of it.
Waymo’s cars, as with Tesla’s and Jaguar’s, use computers built into the car which supply the car’s computers with information about the roads. Necessary information such as how other cars drive and regular obstacles, such as pedestrians, cyclists, and animals.
As we saw earlier, Tesla’s Autopilot does not respond well to construction zones. There are far too many variables and factors that the car is not ready to handle it alone without human intervention. This is in part because millions of miles of driving data are needed so that companies like Tesla and Waymo can collect and aggregate that data.
Because it is difficult to collect that kind of data, engineers train AI with simulation data. The hope is that this training data will be correctly generalized and then become applicable in real life.
Now, this data has improved significantly over the past decade. AI has made major advances in translation, speech generation, computer vision, and object recognition. So it is no longer difficult for AI to identify dogs or bicycles.
Unfortunately for Tesla, Waymo is well in the lead when it comes to miles (or kilometers) driven. Waymo has reported that their vehicles have 20 million miles of overall driving, and 0.09 disengagements every 1,000 miles.
Following Waymo is General Motors’ Cruise, with nearly 500,000 miles and 0.19 disengagements per 1,000 miles. Tesla is not quite in the running yet.
Human Drivers V. Machine Drivers
There is lots of debate over whether humans or machines make better drivers. And the fact of the matter is, there isn’t enough data to prove one way or another. A statistic taken from Vox, explains that human driving produces one fatal accident in every 100 million miles driven.
However, Waymo, the leader in test miles driven, has only just reached 20 million miles driven. And, human driving causes a fatal accident involving a vehicle every 100 million miles driven, while Waymo has not had any fatalities – although Uber has (Elizabeth Herzberg).
One important piece to take from that is the fact that 20 million miles of driving simply isn’t a large enough data pool to draw on.
As we have seen with Tesla’s autopilot, self-driving cars still rely a great deal on human intervention. The “autopilot driver” needs to remain fully attentive to the roads and intervein to stop for lights and signs, as well as some unexpected objects, which may in fact be people.
One such failure of the current design of autonomous cars is that they are still built to assume that people will follow traffic laws. For example, pedestrians will always use crosswalks appropriately. The reason that the Uber failed to recognize Herzberg was that she was walking her bicycle. The AI was not able to successfully identify her. Instead, the AI hit Herzberg at a deadly speed, even after recognizing the object.
Given that Tesla does not have a fully autonomous car, or a cryptocurrency, what does this mean for the near future of the company?
Even though Tesla did not deliver on its fully autonomous car, in 2019, it outsold Mercedes and BMW in terms of luxury vehicles.
As you can see based on the above chart, Tesla’s value continues to increase. Some believe that it will continue to grow steadily. This is in part because while Tesla may have started out slow, deliveries rose from about 2k cars in 2012 to 76k in 2016 and 367k in 2019.
Here is a summary of why Forbes thinks that Tesla’s value is only going to increase:
- Deliveries: It is estimated that deliveries are likely to grow to close to 500k units this year. It is possible for deliveries to stand at over 2 million cars by 2025 and 4 million cars by 2030.
- Revenues: If it is able to increase deliveries as detailed above, Tesla’s Total Revenues could potentially grow from $28 billion in 2020 to $112 billion in 2025 and $212 billion in 2030.
- Net Income: The strong revenue growth will need to be accompanied by steady improvement in margins. Tesla’s Net Margins would need to grow to 13.5% in 2025 and 15% by 2030.
- Tesla’s Valuation: Forbes writers believe that Tesla stock could be worth close to $2,000 by 2025 and more than $3,500 by 2030.
Is it Time for Tesla Currency?
Is it time for a Tesla cryptocurrency? The above sales projections mean that Tesla will be able to improve on its data collection from miles driven. However, it does not address the issue that could incentivize some drivers to more actively participate in sharing data.
Even with more vehicles on the road, the demographic for the testing studies will be very limited. Basically, only those who can afford to buy a luxury vehicle. That means that data will be drawn based on certain kinds of roads, and in some cases, much lower levels of traffic.
Perhaps by incentivizing sales with cryptocurrency, Tesla would draw in customers that were already considering buying a luxury car, just not necessarily a Tesla. That is, they would be able to compete with Jaguar’s model of collecting data and rewarding drivers appropriately.
That’s just my two cents.