Autonomous Car Primer

JAN 2 2018 BY STAFF 17

Some love the idea, some loathe it, but autonomous cars are coming.

Autonomous cars have become a hot topic of discussion in recent years. Established car manufacturers are flirting with the idea of self-driving cars, businesses with no car-making history are also getting in on the act, as are entirely new companies established with the sole goal of producing vehicles that require no driver input.

It may seem like a new and novel technology but the idea of autonomous cars is far from new. First demonstrations of the technology came in the 1980s, and the idea was even explored decades earlier with radio-controlled cars in the United States.


In 1995, a prototype from Carnegie Mellon University’s Navlab program, one of the earliest proponents of autonomous vehicle technology, completed 2,797 autonomous miles during a 2,849 mile journey between Pittsburgh, PA and San Diego, CA.

Autonomous cars were under development for much of the 20th century, but it was the late 2000s and 2010s when the technology really broke into the public eye. American technology giant Google began developing a self-driving car in 2010 ̶ their first foray into the motor industry. The German “big three,” Mercedes-Benz, BMW, and Audi, as well as the wider Volkswagen Group and the likes of Nissan, Volvo, Ford, General Motors, FCA, and PSA, soon followed suit in pushing ahead with their developments.

Along with Google, Apple and Uber also began developing autonomous car technology, despite never building cars of their own before. Uber’s plans involve ditching their controversial driven taxi business model for a huge fleet of autonomous taxis.

What actually is an autonomous car? The Society of Automotive Engineers defines six different levels of autonomy ranging from 0-5.

0 No Automation
1 Drive Assistance (The car assists with things such as accelerating and braking)
2 Partial Automation (The automation of the car is driving-mode-specific)
3 Conditional Automation (Again, mode-specific, but this time instead of the car being primarily driver controlled, the driver will only intervene when absolutely necessary)
4 High Automation (The car is automated, but there can still be human intervention)
5 Full Automation


GM’s Self-Driving Chevy Bolt (Photo by Steve Fecht for General Motors)

Things like radar-guided cruise control and lane assist are quickly becoming commonplace in new cars, and full automation is a distinct possibility in some models although legislation in most regions prevents it at the moment.

However, California, Nevada, Michigan, and Florida has already passed laws permitting autonomous cars that has allowed them to begin testing in those states, as has Germany, the Netherlands, and Spain, while a number of cities across Europe are planning on making changes to accommodate self-driving vehicles.

In the UK, in April plans were unveiled to begin testing driverless cars on motorways from 2019. The plan was backed by the government, who has pledged £100 million (about $132M) as it is “determined that Britain leads the way globally in embracing the safe development of driverless technology”. The Highway Code will also be altered to factor in driverless cars.

The UK Government expects autonomous cars to be on the road “any time from the mid-2020s onwards,” meaning what was once a radical, and relatively far-fetched idea, is now less than a decade away from becoming the norm.

Tesla Autopilot

On the face of it, the idea of driverless cars is a convenient solution for the vast majority of motorists. Most drivers drive out of necessity rather than desire. Still, despite the technology’s best intentions, it isn’t without its flaws, many of which have been thrust into the spotlight.

Tesla Autopilot

Tesla Autopilot 2.0 Update

One such feature that had its fair share of bad press is Tesla’s Autopilot feature. Tesla, pioneers in forward-thinking automotive developments, first offered in 2014 as a driver-assist option with a 90-mph speed limiter, Traffic-Aware Cruise Control, and auto steer, lane change, and parking features. Full driving capability was added with a 2016 update, although its accessibility remains limited for the time being.

Tesla intends Autopilot to eventually be a fully autonomous feature that will be available on all of their products from 2019. The Californian firm acknowledges that there are technical and regulatory hurdles to overcome before its full abilities are offered to the public.

Software updates to Tesla’s second-generation Autopilot are set to introduce a number of new features gradually over a period of time.

Like when they introduced their first mass-produced electric car in 2008, Tesla seem to be leading another automotive revolution with Autopilot; and while the company is pushing ahead with plans to have the full technology in the hands of consumers just two short years from now, the reality is that having it as a bullet-proof and dependable feature is something that remains some way off.

Categories: General

Tags: ,

Leave a Reply

17 Comments on "Autonomous Car Primer"

newest oldest most voted
(⌐■_■) Trollnonymous

So the next time I see one of these cars in SF, I’m going to hold my hand up with “The Force” and watch me stop the vehicle……..ROTFLMAO

If previous article involving motorcycle was any indication, autonomous cars will have to be programmed to drive stupid like humans to reduce crashes (ie, block 2 lanes of traffic for extended time when merging to slower lane).

We can have full automation today if we took all humans off the road. Indeed, dedicated autonomous vehicle road is probably the best solution today and tomorrow.

I’m currently working on a comic fan fiction idea about a guy who gets his licenses suspended.

So what he does is he tells everyone that his electric car is a self driving car to avoid getting pulled over.

The trouble is this idea turns to mayhem when someone who has never driven let alone sees a car before gets behind the wheel of it. And the guy tells him not to worry it’s a self driving car but it might malfunction every now and then.

Don’t get it. Why would the guy let anyone drive with his “fake” autonomous car? That would blow his cover, that He actually is driving the thing. And it would most likely ruin his car… But still more plausible than JJ’s Star Trek stories…

You can either drive in traffic on public roads with nobody in the driver’s seat or you can’t. Waymo can, and has been for 3 months. Tesla can’t, and won’t for 2+ years.

Who’s “leading the revolution”?

That depends on what your yardstick is.

If your yardstick is “Which company’s semi-autonomous cars are more advanced, more capable?” then the answer quite clearly is Waymo. No contest.

If your yardstick is “Which company’s cars are actually operating on the road in large numbers, reducing the number of accidents and saving lives every day?” then the answer quite clearly is Tesla. No contest.

Personally, I’ll go with number of lives saved and serious accidents avoided as the yardstick. But that’s just me. 😉

If you go by number of lives saved and accidents avoided you’ll have to pick Toyota or similar who’ve deployed 100x more cars with the same systems (most notably AEB) that led to Tesla’s 40% reduction. Here’s the NHTSA report with the “almost 40%” number: They compare accident rate before Autopilot was installed vs. after. It is NOT based on Autopilot use (note 22, page 10). They do not split out reductions due to Autosteer vs. other industry-standard features like TACC, PMM and, oh year, AEB. AEB, available for over 10 years (Sec. 2.2, page 1) by itself causes 40% reductions (same section, page 3). Unlike Autosteer, AEB is “always on” by default. NHTSA is a big AEB proponent, spearheading efforts to get automakers to put it in every car by 2022. A couple of final notes. The “almost 40%” reduction was relative to Tesla cars without AEB (or TACC, PMM, etc.). NHTSA does not compare Tesla accident rates to other $70k+ vehicles such as Mercedes. Also, the NHTSA data covers the Mobileye-based AP1.0 system. Tesla dropped AEB for a while when they stopped using Mobileye (temporarily losing the Recommended rating from Consumer Reports). We don’t know how Tesla’s… Read more »

So does that mean the future of the car is riding around while ads play on the front windshield? If so, I’m hoping someone else wins the race.

“Who’s “leading the revolution”?”

Waymo won the contest for “I did it first” but they used a much more expensive hardware suite.

“…the company [Tesla] is pushing ahead with plans to have the full technology in the hands of consumers just two short years from now, the reality is that having it as a bullet-proof and dependable feature is something that remains some way off.”

The reality is that Tesla has been promising full self-drive “in two years” for some time now; I think they are past the one-year mark on that? Let’s hope “full self-driving is only two years away” won’t become a running joke, such as: “Commercial fusion power is 20 years away… and has been for the past 60 years!”

I can’t see ever owning a L5 automation vehicle. There are always scenarios where a computer can’t decide the best course of action. Take a protest turned angry mob for instance. There could be a situation where you need to run people over to escape a violent hoard.

And like with guns, there are likely 100 incidents of Bad Guys missusing the Run Over feature to 1 incident where it actually helps someone escape Bad Guys.

Only 100?

There is an error in the decription of level 4 and level 5.

Level 4 will include nil steering wheel, geo fenced cars. For example taxi fleets are expected to go this way. Expected actors include waymo etc.

Level 5 will include cars with steering wheels, but are not geofenced for example premium cars like porsche audi

From the article: “Tesla intends Autopilot to eventually be a fully autonomous feature that will be available on all of their products from 2019. The Californian firm acknowledges that there are technical and regulatory hurdles to overcome before its full abilities are offered to the public” Tesla Model 3, S, or X vehicles with Autopilot software and current-generation hardware, regardless of software updates, will not meet Level 5 technical and regulatory standards by 2019. The entire vehicle has be be designed and built for full autonomy from its very core on up. The “regulatory and technical hurtles” Tesla must address first are insurmountable without a fundamentally-different vehicle base at a much, much higher cost. The Forbes article link below describes what is required to bring a vehicle to Level 5 Autonomy. Anything less is, as my skateboarding sons call it, “posing”. From the Forbes article: “Now look forward a few years when we may have highly automated vehicles operating in mobility services picking up passengers and dropping them off with no human operator. Imagine the same scenario where the steering actuator fails or the brake apply system and there aren’t even any controls available for a passenger to use to… Read more »

What happens now when one of those non-redundant systems fails? Either we, as humans, crash the car or we know how to handle the situation. If there is a rule on how to handle that situation then a computer can be programmed with the rule.
Eg: brake failure, try the hand brake, if that does not work then turn off the motor so the car naturally slows down. In a manual you can also put it into lower gear. Pull over to the side of the road so you can come to a stop safely.
Unless you’re going to put in a lot of redundancy, which 99% of the time the real world driving shows isn’t required, then there will always be situations where the car can become a safety issue.
As we see with stability control, the computer can apply these rules faster, more accurately and consistently than a person.
I don’t think these articles are logical based on current, inefficient and emotion filled humans experience where we drive all the time and really have few crashed relative to how many cars are on the roads.