web3 levels the playing field for self-driving cars

For decades, self-driving cars have been the stuff of science fiction movies, but now they are equipped with all kinds of sensors, chips, and software. Well, it still is.

This doesn’t mean the industry isn’t evolving. Exciting new initiatives and events are making headlines. But even some trailblazers are questioning this progress, and the stories beneath these headlines feature all-too-familiar headliners: Google’s Waymo, Apple, and General Motors. In a market as disruptive as this, we could and should expect there to be more real disruptors.

The truth is that some of the underlying technologies of the driverless car stack heavily favor centralization and large centralized players. Or so it seems at first glance.

Obviously, taping a camera to a car won’t magically teach it to drive, but neither will connecting that camera to an onboard computer. As far as computers are concerned, a camera feed is just another data stream. The human brain has a complex system of neural connections that extract actionable insights from visual signals, and computers need something similar. You need your own vision.

Computer vision is a subfield of the broader artificial intelligence (AI) industry, or more accurately machine learning (ML), that allows driverless cars to “see” the world around them. AI algorithms are often used to process other sensor feeds, such as LiDAR, to improve a vehicle’s ability to navigate physical spaces. And the problem with such models is that they require incredibly large amounts of data to train.

Companies, wary of how far they can get with simulated data sets, have long struggled to acquire real data to train their models. The driverless car industry is no exception. Companies can use simulations, much like those found in video games, to record different scenarios and bootstrap data sets, but they can only go so far. From weather conditions to regional characteristics, real-world data is critical to making self-driving cars safe and reliable. That’s why San Francisco residents can see driverless taxis cruising for hours without passengers. They are not looking for passengers. They are collecting data.

The challenge of collecting data sets of sufficient size and quality at a pace to continue doing business is a hurdle for the autonomous vehicle industry. This keeps the playing field uneven and leads to a shift towards large, centralized entities. Large, centralized companies collect enormous amounts of data, while new companies face data challenges that hinder their progress. This is casting the shadow of an oligopoly over a nascent and promising market, and we all know what this means for the common man.

The solution is already available to thousands of vehicles on the roads of every city and country every day. Most capture large amounts of data while on the move, and with the right incentives, drivers are likely to do the labeling themselves. Take a look at CAPTCHA. Testing pedestrians, motorcycles, and traffic lights are all data labeling exercises that people do simply to access a website or service.

Accumulating all this data into a massive set provides both emerging startups and enterprises with all the real-world learning material their models might need. These data sets can be as varied or location-specific as needed based on real-world scenarios, conditions, and details. But to unlock access to data, the industry first needs an entirely new data paradigm.

This paradigm should leverage blockchain as a shared, vendor-neutral core infrastructure and transaction layer to prevent the rise of yet another siled ecosystem. We also need to return control of their data and privacy to drivers by leveraging autonomous data and identity for both drivers and vehicles.

Sovereign Identity operates as a web3 wallet that stores cryptographic proofs of various user attributes issued by trusted entities such as authorities or car manufacturers. Data consumers can use this to see what data sellers (in this case drivers) can select for sale. With both web2 and web3 companies already developing blockchain-based mobility infrastructure, such as Gaia-X moveID in Europe, the prospect is not as far-fetched as one might think.

This self-sovereign data paradigm will transform drivers into active stakeholders in the digital mobility space, allowing them to monetize the data they generate during their daily commute. It will also address dataset issues across the autonomous vehicle industry, providing much-needed support to the industry by providing equal access to a raw data sharing marketplace for all participants.

Despite all the promise, truly self-driving cars remain elusive. This is partly due to the difficulties associated with collecting data sets to train AI models that would help drive such cars. Embracing the web3 data paradigm is the industry’s best opportunity to unlock access to a virtually unlimited pool of training data while maintaining a spirit of healthy open competition.

This article is co-authored. Sheridan JonesCo-founder of Ocean Protocol Leonard DollöhterCo-founder of Peak.

Follow us on Google News