Silicon Valley and the electric / autonomous car

Jeroen · 15th October 2015, 22:22

I am fortunate in the sense that my employer has an excellent training program which allows me to participate in some very interesting courses.

Last week I was at Stanford University, Palo Alto, in the heart of Silicon Valley. The course was all about innovation and entrepreneurship.

We spend a whole morning at the CARS lab, with Prof Christiaan Gerdes. See http://cars.stanford.edu

A pretty cool place for any petrol head. Now the course had nothing to do with the car industry as such, more about innovation and entrepeneurship in general and in the context of Silicon Valley. Even so, we did discuss a bit about the technology involved as well. I certainly learned some new things, so I though I would share some of the more remarkable things I heard/learned.

One of the things, that is preached if not to say is a real mantra all around Sillicon valley is the need for very early prototyping. No matter what it is you are thinking of producing. Its all about mission orientation and the earlier you can visualise, see and feel the better your end product is likely to be. Learn by doing as they call it

So quite some years ago Nissan came to Stanford with a question on the ‘steer by wire” they (Nissan) at the time was developing. Their question was, how can we diagnose faults from very early on and thus reduce the need for redundancy.

Rather then to tackle this problem purely from the ‘steer by wire’ perspective Stanford convinced them to built a complete prototype car. When Nissan asked them to what purpose, the answer was: we don’t know, but we are very likely to learn an awful lot. Nissan did agree and they built a first prototype car with independent front steering and independent rear drive. As they started driving this contraption it became obvious that as the steering and suspension handles all the forces coming from the tires, they could also use this to calculate for instance road friction.

Long story short, this ultimately became the “Directive Adaptive Steering on the 2014 Infiniti Q50 in 2014. That by the way was quite a few years after Stanford concluded their research.

Another story they told was around Shelley. http://news.stanford.edu/news/2010/f...ak-020310.html

The car, with different stickers, still sits in their garage

Silicon Valley and the electric / autonomous car-img_43301.jpg

Shelley is all about Stanford involvement in autonomous cars and I refer to the above link for some more information. But this car can really move. They are able to race it, without driver, close to amateur race drivers times on tracks.

I had a long conversation with one of the graduates who is looking after the positioning system such as GPS, inertial reference system and such. Very interesting.

What the Shelley Pike Peak did not tell was that initially Audi was very reluctant to even acknowledge that they had driven up the peak autonomously. The reason was that during what was supposed to be the last climb, the pursuit helicopter that was also filming crashed. The helicopter was totalled, all occupants escaped more or less unharmed, but Audi just did not want this out in the public domain, so essentially the whole project was shelved. Only considerable time later, Audi came back and Shelley made an appearance on stage at one of the international car shows as the shining example of Audi’s innovative inroad in autonomous cars. At that show the then Audi CEO mentioned the Peak climb for the very first time, in public.

They also have their own driving simulation and I spend some time in it.

Silicon Valley and the electric / autonomous car-img_43395.jpg

It is a static simulator and I found quite disorientating at first. When it comes to autonomous cars there are, broadly speaking, two different approaches. There is the Google approach which is you get in the car, you push a button and that’s it. you don’t have to do anything else. The other approach and it seems most current car manufacturers have taken this approach is that there is constant need for driver and computer interaction and handing over of command. So when the computer gets confused he tells you to take control again. A lot of research around this handing over of control was done in this simulator. They found it took on average 3.5 - 5.5 seconds for a driver to re-engage. Imagine, you are driving and your car is autonomous mode. You are maybe looking at your smart phone and the car tells you to take control. It would take you that long to fully understand what is happening, absorb the environment etc and be in actual control of the car. At a speed of around 100km/h that also means that the sensors must have the ability to see ahead for several hundred meters!

Tesla has just launched a new software version for their cars, that apparently allows for certain autonomous driving, although I understand you are not supposed to let go of the steering wheel!

Here is another interesting angle to the software updates of Tesla. Tesla owners actually welcome these software updates. They don’t see it as bug fixers or that the car was initially not well designed. They see it at as a proof point of the continuous innovation of Tesla and are very happy to upgrade the software of their cars at regular intervals. There are other examples, in other industries (media) that have been successful in this approach, sort of continuous evolution of their product. Very often based on input from consumers. Again learning by doing.

Nobody I spoke to wanted to make any prediction about when we will see full autonomous cars on the road. However, they all pointed out that the typical time between research at Stanford and when that researched actually found its way into products to the market was at least 10 years!

Here are a few more shots of their garage.
For me personally, with a huge interest in tools a bit of a wet dream.

Silicon Valley and the electric / autonomous car-img_43312.jpg

Here is another prototype for an electric car with independent, electric steering.

Silicon Valley and the electric / autonomous car-img_43323.jpg

When you say Silicon Valley, Stanford University, entrepreneur and innovation you have to mention Elon Musk of course. We were lucky as Elon was at the campus being interview by Venture Capitalist Steve Jurvetson. Interesting conversation. Later we met with Steve and discussed the world of VC for several hours.

Silicon Valley and the electric / autonomous car-img_43406.jpg

You might remember the original Back to the future movie? As the real back to the movie addict will know Doc travel to the future, October 8th, 2015 to be precise. So obviously a DeLoran materialised on campus, and Tesla put one of their new models on display as well. Nice little party.

Silicon Valley and the electric / autonomous car-img_43447.jpg

All in all, spending a week in Silicon Valley is a remarkable experience, no matter what.

Jeroen

Rehaan · 17th October 2015, 13:36

Interesting experience Jeroen - thanks for sharing!

Here's a video of the Autonomous TT doing pikes peak:

https://www.youtube.com/watch?v=IFwIlflmk2Y

Naturally, the TT video did have a bit of a 'climb dance' feel to it at the start!

Though once the video started, it was clear that it wasn't exactly on-the-limit driving. Still impressive though.

It did the 12.42 (19.99km) course in about 27 minutes.

Just for reference, Wikipedia tells me that the current course record is 8 min 13 sec (Sebastien Loeb - 2013), though it's very interesting to also see the amount of electric cars setting 9-minute times too.

[OT question: How come Loeb's time is SO much faster than everyone else, including Monster Tajima, who held the record for 6 consecutive years? Is it a question of equipment and backing, or was it weather conditions, or an altered course/surface, or something else?]

Quote:

Originally Posted by Jeroen

the need for very early prototyping

This is becoming easier and easier every year with rapid prototyping becoming so much more advanced and affordable.

Jeroen · 17th October 2015, 15:41

Quote:

Originally Posted by Rehaan

Just for reference, Wikipedia tells me that the current course record is 8 min 13 sec (Sebastien Loeb - 2013), though it's very interesting to also see the amount of electric cars setting 9-minute times too.

[OT question: How come Loeb's time is SO much faster than everyone else, including Monster Tajima, who held the record for 6 consecutive years? Is it a question of equipment and backing, or was it weather conditions, or an altered course/surface, or something else?]

Thanks. The Peak Pike climb was in 2010 and since they have raced it on various circuits and improved it to get close to lap time of the top amateur drivers. Do bear in mind that it takes a lot of preparation. This can not be compared to a regular autonomous car on the public road. For instance they use very accurate GPS (better than 1 cm) which requires additional ground base stations. They also need to set the car up specifically for the circuit, temperature, humidity etc. That all takes time.

This is where the big difference between autonomous car and car driven by humans come into play. Humans are very good at adapting to the environment around them. It comes to a large extend naturally. Some racing drivers are phenomally good at it. I can’t really commit on the names you mention, because I don’t really follow modern racing much. But take a driver such as Stirling Moss and stick him in a car, any car and he will be driving it on the very limited almost immediately. Currently it takes endless tweaking and tuning to get a computer to do so. And even then, currently, they can’t match human performance.

But also other events. For instance recognising a little old stooped over lady with a stick who might be crossing the road, slowly, 200 meters in front of the car. Humans have no problem seeing that and anticipating (if you have been taught to drive properly). Same with kids playing with a ball on the pavement. You know that ball might bounce into the road and the kid in hot pursuit. It comes naturally but a computer has to be learned/programmed to adapt to this incredibly complicated environment.

They talked us through several of these examples and it is one of the things, in combination with what I road in my original post around how far the sensors need to see ahead that autonomous cars, apparently, are still struggling with.

Jeroen

mohaks19 · 18th October 2015, 11:32

Well, this post is truly ecstatic and VERY informative for me, particularly because I'm applying to Stanford University for undergraduate admissions this year. You're one lucky man to have visited their labs! Stanford is particularly proud about Shelley and rightly so. By the way, did you visit Volkswagen Automotive Innovation Lab (VAIL) at Stanford? It's lovely and well, truly a wet dream for someone like me. It's fully funded by Volkswagen AG, employed particularly for emissions, autonomous research and pedestrian safety systems (Driver Assist and the likes). In fact, Audi's complete 'Piloted Driving' technology has been developed in collaboration with Stanford University. Anyways, thanks for this wonderful post!

Jeroen · 18th October 2015, 13:31

Thanks. No I did not visit VAIL. But I'm going back to Stanford for another week in November. Who knows. Good luck with your application. It is hugely competitive. They only admit a few thousands students every year and several tens of thousands apply! So it's all down to your academic credentials and a few unmentionables. The unmentionables are to do with campus diversity. So if you put on your application form you are an Gay Indian raised by a Lesbian couple from Nigeria and you lived in the slums of Bombay for most of your life you will go straight to the top of the list. If not better make sure you have the best academic credentials ever, because you are competing with some ultra smart people. But if you make it you are absolutely going to love it! Guaranteed!

blackwasp · 19th October 2015, 18:08

Wow, this is proper drool worthy. Thanks for sharing this experience. This might be the motivation needed for me. My interests are in electronics and automotive fields. This looks like the perfect place to do/make/create projects.

Mr. Musk is one of my idols and nice to know you had the opportunity to hear him.

Udayrg · 20th October 2015, 12:13

A fascinating visit it is..... Jeroen!!!!. You might be interested in reading about the DARPA challenge 2005 (Defense Advanced Research Projects Agency). The vehicle from Stanford university built with assistance from Volkswagen Electronics Research Laboratory (ERL) won the race across Nevada Desert. It completed staggering 212 kms distance autonomously in less than 7 hours. You might find more details if you google it. In my eng. bachelor days I used to follow Telematics coined from Telecommunication and information systems. In-fact had presented a student level paper in competition on Telematics. And put forth many things from the Stanley vehicle and my concepts of passive and active control systems. Had two things in my mind during Eng. as a career---- Telematics OR CFD ( computational Fluid Dynamics). Finally did land in CFD. Well a topic close to my heart is plugged in to the discussion. Let us keep it posted and updated......

Jomz · 31st October 2015, 18:24

Very interesting. hope you enjoyed.

Jeroen · 31st October 2015, 18:39

Quote:

Originally Posted by Jomz

Very interesting. hope you enjoyed.

thanks, very much. I will be back first week of December for another week. Looking forward to it, apart from getting there. Bit of drag, long flights, lots of time difference.

Jeroen

15th October 2015, 22:22	#1
Jeroen Distinguished - BHPian Join Date: Oct 2012 Location: Delhi Posts: 8,101 Thanked: 50,880 Times View My Garage	Silicon Valley and the electric / autonomous car I am fortunate in the sense that my employer has an excellent training program which allows me to participate in some very interesting courses. Last week I was at Stanford University, Palo Alto, in the heart of Silicon Valley. The course was all about innovation and entrepreneurship. We spend a whole morning at the CARS lab, with Prof Christiaan Gerdes. See http://cars.stanford.edu A pretty cool place for any petrol head. Now the course had nothing to do with the car industry as such, more about innovation and entrepeneurship in general and in the context of Silicon Valley. Even so, we did discuss a bit about the technology involved as well. I certainly learned some new things, so I though I would share some of the more remarkable things I heard/learned. One of the things, that is preached if not to say is a real mantra all around Sillicon valley is the need for very early prototyping. No matter what it is you are thinking of producing. Its all about mission orientation and the earlier you can visualise, see and feel the better your end product is likely to be. Learn by doing as they call it So quite some years ago Nissan came to Stanford with a question on the ‘steer by wire” they (Nissan) at the time was developing. Their question was, how can we diagnose faults from very early on and thus reduce the need for redundancy. Rather then to tackle this problem purely from the ‘steer by wire’ perspective Stanford convinced them to built a complete prototype car. When Nissan asked them to what purpose, the answer was: we don’t know, but we are very likely to learn an awful lot. Nissan did agree and they built a first prototype car with independent front steering and independent rear drive. As they started driving this contraption it became obvious that as the steering and suspension handles all the forces coming from the tires, they could also use this to calculate for instance road friction. Long story short, this ultimately became the “Directive Adaptive Steering on the 2014 Infiniti Q50 in 2014. That by the way was quite a few years after Stanford concluded their research. Another story they told was around Shelley. http://news.stanford.edu/news/2010/f...ak-020310.html The car, with different stickers, still sits in their garage Shelley is all about Stanford involvement in autonomous cars and I refer to the above link for some more information. But this car can really move. They are able to race it, without driver, close to amateur race drivers times on tracks. I had a long conversation with one of the graduates who is looking after the positioning system such as GPS, inertial reference system and such. Very interesting. What the Shelley Pike Peak did not tell was that initially Audi was very reluctant to even acknowledge that they had driven up the peak autonomously. The reason was that during what was supposed to be the last climb, the pursuit helicopter that was also filming crashed. The helicopter was totalled, all occupants escaped more or less unharmed, but Audi just did not want this out in the public domain, so essentially the whole project was shelved. Only considerable time later, Audi came back and Shelley made an appearance on stage at one of the international car shows as the shining example of Audi’s innovative inroad in autonomous cars. At that show the then Audi CEO mentioned the Peak climb for the very first time, in public. They also have their own driving simulation and I spend some time in it. It is a static simulator and I found quite disorientating at first. When it comes to autonomous cars there are, broadly speaking, two different approaches. There is the Google approach which is you get in the car, you push a button and that’s it. you don’t have to do anything else. The other approach and it seems most current car manufacturers have taken this approach is that there is constant need for driver and computer interaction and handing over of command. So when the computer gets confused he tells you to take control again. A lot of research around this handing over of control was done in this simulator. They found it took on average 3.5 - 5.5 seconds for a driver to re-engage. Imagine, you are driving and your car is autonomous mode. You are maybe looking at your smart phone and the car tells you to take control. It would take you that long to fully understand what is happening, absorb the environment etc and be in actual control of the car. At a speed of around 100km/h that also means that the sensors must have the ability to see ahead for several hundred meters! Tesla has just launched a new software version for their cars, that apparently allows for certain autonomous driving, although I understand you are not supposed to let go of the steering wheel! Here is another interesting angle to the software updates of Tesla. Tesla owners actually welcome these software updates. They don’t see it as bug fixers or that the car was initially not well designed. They see it at as a proof point of the continuous innovation of Tesla and are very happy to upgrade the software of their cars at regular intervals. There are other examples, in other industries (media) that have been successful in this approach, sort of continuous evolution of their product. Very often based on input from consumers. Again learning by doing. Nobody I spoke to wanted to make any prediction about when we will see full autonomous cars on the road. However, they all pointed out that the typical time between research at Stanford and when that researched actually found its way into products to the market was at least 10 years! Here are a few more shots of their garage. For me personally, with a huge interest in tools a bit of a wet dream. Here is another prototype for an electric car with independent, electric steering. When you say Silicon Valley, Stanford University, entrepreneur and innovation you have to mention Elon Musk of course. We were lucky as Elon was at the campus being interview by Venture Capitalist Steve Jurvetson. Interesting conversation. Later we met with Steve and discussed the world of VC for several hours. You might remember the original Back to the future movie? As the real back to the movie addict will know Doc travel to the future, October 8th, 2015 to be precise. So obviously a DeLoran materialised on campus, and Tesla put one of their new models on display as well. Nice little party. All in all, spending a week in Silicon Valley is a remarkable experience, no matter what. Jeroen Attached Thumbnails
	(21) Thanks

18th October 2015, 11:32	#4
mohaks19 BHPian Join Date: Aug 2015 Location: New York Posts: 70 Thanked: 116 Times	Re: Silicon Valley and the electric / autonomous car Well, this post is truly ecstatic and VERY informative for me, particularly because I'm applying to Stanford University for undergraduate admissions this year. You're one lucky man to have visited their labs! Stanford is particularly proud about Shelley and rightly so. By the way, did you visit Volkswagen Automotive Innovation Lab (VAIL) at Stanford? It's lovely and well, truly a wet dream for someone like me. It's fully funded by Volkswagen AG, employed particularly for emissions, autonomous research and pedestrian safety systems (Driver Assist and the likes). In fact, Audi's complete 'Piloted Driving' technology has been developed in collaboration with Stanford University. Anyways, thanks for this wonderful post!
	() Thanks

18th October 2015, 13:31	#5
Jeroen Distinguished - BHPian Join Date: Oct 2012 Location: Delhi Posts: 8,101 Thanked: 50,880 Times View My Garage	Thanks. No I did not visit VAIL. But I'm going back to Stanford for another week in November. Who knows. Good luck with your application. It is hugely competitive. They only admit a few thousands students every year and several tens of thousands apply! So it's all down to your academic credentials and a few unmentionables. The unmentionables are to do with campus diversity. So if you put on your application form you are an Gay Indian raised by a Lesbian couple from Nigeria and you lived in the slums of Bombay for most of your life you will go straight to the top of the list. If not better make sure you have the best academic credentials ever, because you are competing with some ultra smart people. But if you make it you are absolutely going to love it! Guaranteed!
	(2) Thanks

19th October 2015, 18:08	#6
blackwasp Senior - BHPian Join Date: Apr 2015 Location: Navi Mumbai Posts: 2,974 Thanked: 26,325 Times View My Garage	Re: Silicon Valley and the electric / autonomous car Wow, this is proper drool worthy. Thanks for sharing this experience. This might be the motivation needed for me. My interests are in electronics and automotive fields. This looks like the perfect place to do/make/create projects. Mr. Musk is one of my idols and nice to know you had the opportunity to hear him.
	() Thanks

20th October 2015, 12:13	#7
Udayrg Newbie Join Date: Apr 2010 Location: Coimbatore Posts: 8 Thanked: 0 Times	Re: Silicon Valley and the electric / autonomous car A fascinating visit it is..... Jeroen!!!!. You might be interested in reading about the DARPA challenge 2005 (Defense Advanced Research Projects Agency). The vehicle from Stanford university built with assistance from Volkswagen Electronics Research Laboratory (ERL) won the race across Nevada Desert. It completed staggering 212 kms distance autonomously in less than 7 hours. You might find more details if you google it. In my eng. bachelor days I used to follow Telematics coined from Telecommunication and information systems. In-fact had presented a student level paper in competition on Telematics. And put forth many things from the Stanley vehicle and my concepts of passive and active control systems. Had two things in my mind during Eng. as a career---- Telematics OR CFD ( computational Fluid Dynamics). Finally did land in CFD. Well a topic close to my heart is plugged in to the discussion. Let us keep it posted and updated......
	() Thanks

31st October 2015, 18:24	#8
Jomz BHPian Join Date: Sep 2006 Location: Detroit, MI,USA Posts: 834 Thanked: 335 Times	Re: Silicon Valley and the electric / autonomous car Very interesting. hope you enjoyed.
	() Thanks