Technology To Bring Self-Driving Capabilities To Older Vehicles

Retrofitting existing cars could jump-start (figuratively) the self-driving car market.

It would also reduce the initial expense for tech companies, making them more competitive with the OEMs (original equipment manufacturers, eg, the automakers). JL

Jonathan Gitlin reports in ars technica:

An automotive renaissance won't be that impactful for current cars unless it can be retrofitted to the 300 million cars on the road. Knowing how to talk to different OEM vehicles is half the battle. An ecosystem of companies exists ready to give you the tools to take out a conservative engine map for something that makes greater use of your engine's capabilities. And business fleets, too, although they're looking for efficiency gains, not a faster zero-to-60 time.
Truly, we are living in the age of the software-defined vehicle. Beginning with systems like electronic fuel ignition and anti-lock brakes, black boxes and ECUs have proliferated throughout our vehicles, all networked together on the Controller Area Network (CAN) bus. Standardized back in the olden days, when even cell phones were uncommon, the CAN bus even provides a handy way to connect devices that can query other ECUs on the car's internal network.
Being able to read information from that network was originally envisioned as a way to make servicing easier—hence the reason the port is known as an OBD (on-board diagnostics) II port. But these days, it's far more likely to be pressed into service as a way of enabling some form of telematics, whether that's a device that records your driving style for your insurance company in return for a discounted rate or gadgets like Automatic, Mojio or Verizon's Hum that let you monitor your car via a mobile app.
But the OBDII port can do more than just read data. Developed back in the days before anyone envisioned how the Internet would be transformed by wireless modems, it also allows write-access to devices on the network—something that seems much more problematic in the era of the vehicle hack. (OBDII's designers did take the idea of malicious use of the port into account; it's always in a spot where the driver can reach and unplug it, even though its creators assumed it wouldn't be connected to the world outside.)
Leaving aside the problem of car hacking for another day, the ability to write data to controllers on a car can actually be quite useful—like the idea of reflashing an engine ECU to get better performance, for instance. Over on the enthusiast side of the market, an entire ecosystem of companies exists ready to give you the tools to take your WRX, GT-R, or whatever to the next level. Those companies do so by swapping out a conservative engine map for something that makes greater use of your engine's capabilities. And business fleets do it, too, although they're looking for efficiency gains, not a faster zero-to-60 time.

Enter Derive Systems

A company called Derive Systems has been playing in this space for some time now, both on the enthusiast side of things and also with fleets. After all, if you're Comcast and have thousands of trucks that spend their lives visiting install sites, remapping your engines—at the expense of, say, towing performance—can save between 8 and 12 percent on fuel costs, according to Derive's CEO, David Thawley. Thawley told us that more than 1.5 million installs of his company's technology platform are currently driving around. As a result, the company has built up a hefty knowledge base of how to work with different OEM systems, many of which feature different implementations of the field's standards. And Thawley now wants to offer that know-how to new players in the connected-car world so they can get up to speed.
Thawley told Ars:

The way I think about it is we're at the point of an automotive renaissance. But the reality is it won't be that impactful for current cars unless it can be retrofitted to the 300 million cars on the road. So we can say "Hey, Geohot, as you're creating self-driving tech, if you want to enable it through write-access and not try and do it in a one-by-one business-development process with OEMs that are tremendously difficult to work with, we'll provide this one-stop shop where you have the tools—the pick and shovels—to be successful in this gold rush."

(Geohot—real name George Hotz—has a company, comma.ai, that is developing an aftermarket semi-autonomous driving system, but it currently only works with a single Honda model and a single Acura model.)
Another use case Thawley offered was working with a company that builds Bluetooth blood-alcohol detectors. "It's more compelling when combined with our technology to prevent the vehicle from starting if the detector shows you're over the limit," he said.
Of course, plenty of people out there will be wary of such a powerful tool in this age of data breaches. But Hawley thought that shouldn't be an issue, provided those using his tools act responsibly. "Anything with connectivity outside the vehicle needs to follow industry best practices, and you have to be diligent," he said.
From the sounds of things, Derive's platform could be an effective shortcut for developers looking to get their tech working across multiple OEMs. I wonder if Hotz will give him a call?