During my recent trans-continental drive, after about hour 300 of slogging across America's infinitely-wide state, Oklahoma, I found myself daydreaming about the possible benefits of an autonomous car. But I wanted to keep this archaic old RV, so I thought what about an autonomous driver-bot? This should be a thing, right?

I've been in autonomous cars before, and so far they've all been converted production models, like a Nissan Leaf, or those Priuses Google has been using. The electronics and sensor packages are discreet units that are installed in/on the cars, and the car's actual controls are usually accessed via the car's drive-by-wire system.

It should be possible, I'd think, once the electronics are refined and packaged efficiently, to create an autonomous driving unit that is transportable and can be easily installed in pretty much any car. This could allow for use cases like the one I experienced, where you could make a vintage car (or RV, why not) into a self-driving (or really, robot-driven) vehicle quickly and easily.


There are, of course, significant challenges and restrictions associated with this. The unit would need to be roughly humanoid in form, to occupy the driving position of the car. Since you can't always count on drive-by-wire controls in a car, especially a vintage one, physical interfaces to the car's controls would be needed.

For the pedals, this problem has already been solved by the auto towing and trailering submarket. There's products known as supplemental braking systems that are essentially robotic feet that press a towed vehicle's brake pedal. This same basic concept would be used on our robotic chauffeur, but instead of just the brake, a unit, connected via Bluetooth or other wireless link to the main torso unit, would actuate the brake, accelerator, and, if present, the clutch.


The pedal unit, since it could be wirelessly controlled, could be repositioned on the vehicle's floor wherever needed. The steering wheel controls on the main body of the robot would need to be adjustable and secured to the wheel via some manner of clamp system. An extra arm on the robot could be trained to find the locations of other controls, like lights, wipers, indicators, etc.

The robot's main body would be placed in the driver's seat and secured with the standard seat belts, if available. If not, supplemental restraints could be used to secure the unit in place. The bottom of the main unit would have a rubberized pad to eliminate sliding on even the most American of huge vinyl bench seats.

If an OBD-II port is available, the robot could connect to that to get vehicle data, like speed, RPMs, etc to assist with driving. If it's an older vehicle, speed would be calculated via the robot's GPS system, and RPMs could be assessed either via audio cues, visual inspection of a tachometer, or some combination of both.

If the car requires shifting, there would likely need to be a third arm dedicated to controlling the gearshift. Initial setup would allow the robot to pick between standard patterns (4-speed H-pattern, 3-on-the-tree, 6-speed, whatever) and perhaps a brief training session would be needed with a driver shifting to train the arm to the gear's locations. Clutch point and pressure would likely require a similar training session to get right as well.

The head of the robot would contain the main visual, sonar, radar, and other sensors, though it's likely additional sensors would need to be mounted to the exterior of the car as well, where they could communicate wirelessly with the main unit.

These autonomous driving units would have to be a bit more flexible and robust than units built into a single car, for the obvious reason that their inputs would have much more variable results than in cars who's characteristics are already known. Still, I think the technology is rapidly approaching a point where this could be possible.

Think how handy this could be! Aside from taking over driving on long, straight, boring stretches of highway, the robot chauffeur could be used as a tow alternative if you have multiple cars to transport to a rally or car show or wherever. Just drive your, say, Lancia Stratos and let the robot follow you in your restored 2CV. Have it take your classic to the detailer while you lay around in your underwear, eating corn chips and playing video games. Start a long trip in your Jensen Interceptor, and then pull it out of the hatch when you want a nap or just get tired of driving. And, perhaps most importantly, perhaps systems like these could be the way to keep older, interesting cars on the road even in areas where autonomous cars may become mandated at some point in the future.

Autonomous cars are coming, and while they're not exactly an enthusiast's dream, there are some pretty huge safety and convenience advantages hey could offer. It seems that the option to take advantage of these benefits, on demand, in any car you want, could be a really useful thing.

So who's got an old robot sitting around looking for work?