World premiere: the pilot-driving Audi A8 The new A8 is the first production automobile to have been developed specially for highly automated driving. The Audi AI traffic jam pilot takes charge of driving in slow-moving traffic at up to 60 km/h (37.3 mph) on freeways and highways where a physical barrier separates the two carriageways. The system is activated using the AI button on the center console.
The traffic jam pilot manages starting, accelerating, steering and braking. The driver no longer needs to monitor the car permanently. They can take their hands off the steering wheel permanently and, depending on the national laws, focus on a different activity that is supported by the car, such as watching the on-board TV. As soon as the system reaches its limits, it calls on the driver to take back control of the task of driving.
From a technical perspective the traffic jam pilot is revolutionary. During piloted driving, a central driver assistance controller (zFAS) now permanently computes an image of the surroundings by merging the sensor data. As well as the radar sensors, a front camera and the ultrasonic sensors, Audi is the first car manufacturer also to use a laser scanner. The introduction of the Audi AI traffic jam pilot means the statutory framework will need to be clarified in each individual market, along with the country-specific definition of the application and testing of the system. The brand’s high quality standards are equally applicable in the realm of highly automated driving. In addition, a range of approval procedures and their corresponding timescales will need to be observed worldwide. Audi will therefore be adopting a step-by-step approach to the introduction of the traffic jam pilot in production models.
The Audi AI remote parking pilot and the Audi AI remote garage pilot autonomously steer the A8 into and out of a parking space or a garage, while the maneuver is monitored by the driver. The driver need not be sitting in the car. They start the appropriate system from their smartphone using the new myAudi app. To monitor the parking maneuver, they hold the Audi AI button pressed to watch a live display from the car’s 360 degree cameras on their device…
Hands off the wheel in a dense, rapidly changing traffic environment … we’ll see if that lasts to release. Tesla has caught enough grief over failures of its system where the driver is supposed to keep hands on the wheel and tries to enforce that, but actively telling people they can take their hands off the wheel and pay attention to something else in a not yet released system sounds like a set up for a retraction.
So, the step by step introduction sounds like preparing for not actually delivering until well past the initial date of introduction of the model.
And, using old technology sounds like it is unlikely to be anything like as “new” as they are trying to make out.
And, using old technology sounds like it is unlikely to be anything like as “new” as they are trying to make out.
While I was a manager in a Fortune 50 company I had an engineer on my staff for a while who had previously worked on an automated commuter light rail project. At the time, this was very new stuff, only in Japan were there any implemented autonomous rail systems.
The engineer who worked for me went to Japan in order to review their technology. Not too far into the review he noticed that they had elected to use mechanical relays rather than solid state switches. He asked why they would chose an older technology for this critical function.
The Japanese design engineers responded (I paraphrase): “We have a long history and a lot of experience with mechanical relays. We have very reliable statistics regarding MTBF, MTTR (mean time between failures, mean time to repair), etc. {I won’t provide the entire list - I don’t remember it and it doesn’t add new info anyway]. We’ve built redundancy and failsafe into every aspect of the entire system. The likelihood of a catastrophic failure resulting in loss of life or even any type of collision can be calculated and is diminishingly small. We can’t make these calculations with confidence for solid state switches.”
The point is that often engineers who design these systems have very good reasons to employ older technologies rather than cutting edge. I’m not asserting that was the rationale used by Audi. I have no knowledge of their design decision process. Just saying older doesn’t always mean inferior.
The Japanese design engineers responded (I paraphrase): “We have a long history and a lot of experience with mechanical relays. We have very reliable statistics regarding MTBF, MTTR (mean time between failures, mean time to repair), etc. {I won’t provide the entire list - I don’t remember it and it doesn’t add new info anyway]. We’ve built redundancy and failsafe into every aspect of the entire system. The likelihood of a catastrophic failure resulting in loss of life or even any type of collision can be calculated and is diminishingly small. We can’t make these calculations with confidence for solid state switches.”
The point is that often engineers who design these systems have very good reasons to employ older technologies rather than cutting edge.
Another point is that AI will allow you to test scenarios in virtual reality. VR models will be used to generate experience and test various scenarios inside the computer generated world. What was not possible to test in a fast and cheap way in Japan at that time will soon be possible. This will allow implementation of newer technologies more quickly.
I’m not asserting that was the rationale used by Audi. I have no knowledge of their design decision process. Just saying older doesn’t always mean inferior.
I would guess that Audi started their project several years ago when the newer products from Nvidia were not yet on the market. Rather than switch to the newer chips it was probably deemed that it would be better to launch sooner (to get into the 2018 models) with a design that’s good enough for the functionality they need right now.
I would guess that Audi started their project several years ago when the newer products from Nvidia were not yet on the market. Rather than switch to the newer chips it was probably deemed that it would be better to launch sooner (to get into the 2018 models) with a design that’s good enough for the functionality they need right now.
I would wager that you are correct. Changing the AI chip at the heart of the functionality they are offering would send them right back to the very beginning of the design, prototype, test cycle. They want to be on the street (where allowable) by 2018 model release. If they swap chips now it’s unlikely they would be ready for product release by the fall of this year.
This just goes to show that until a product is across the chasm, nothing is set in stone.
This said, as I wrote about in the linked to post, the functionality of this autonomous driving is likely to become more annoying than useful once the novelty wears off. The functionality is so limited to circumstances that its usefulness will have to be weighed against the rarity that it will be used and the aggravation it is likely to cause.
E.g. rush hour traffic. It only works on specific types of and specific portions of highways, and then only up to 37 m.p.h. The usual rush hour that I am use to, even in Atlanta (and I think Minneapolis has worse rush hour, but curvier roads (thus making the utility of this function less useful in Minneapolis, even though it might be of more prevalent use) has stop and go traffic, BUT, you stop, move at slow speed, and then your back up to speed for awhile above 37 mph.
This feature will be limited to certain portions of the road, and not others, and will continually kick on and off as traffic stops and goes.
TBD, but seems like it may be more trouble than it is worth, except in specific circumstances.
Minneapolis rush hour can be atrocious and more fit for this technology. But the roads are narrower, they curve more, and I don’t know how much of the highways (particularly given the extensive construction going on all over the highway system currently) would be considered suitable of the system to operate.