Volvo Cars announces production-viable autonomous driving system; targeting limited pilot customer rollout by 2017
19 February 2015
Volvo Cars has developed what it says is a unique, complete system solution that makes it possible to integrate self-driving cars into real traffic. Based on its analysis of potential technical faults, Volvo Cars designed a complete production-viable autonomous driving system. The key is a complex network of sensors, cloud-based positioning systems and intelligent braking and steering technologies.
Volvo Cars’ Autopilot system is designed to be reliable enough to allow the car to take over every aspect of driving in autonomous mode. The technology advances a crucial step beyond the automotive systems demonstrated so far since it includes fault-tolerant systems. With the Drive Me project entering its second year (earlier post), Volvo Cars is thus moving rapidly towards its goal of placing 100 self-driving cars in the hands of customers on selected roads around Gothenburg by 2017.
It is relatively easy to build and demonstrate a self-driving concept vehicle, but if you want to create an impact in the real world, you have to design and produce a complete system that will be safe, robust and affordable for ordinary customers.—Dr. Erik Coelingh, Technical Specialist at Volvo Cars
The main challenge, Volvo said, is to design an Autopilot that is robust for traffic scenarios as well as for technical faults that may occur. It cannot be expected that the driver is ready to suddenly intervene in a critical situation. Initially, the cars will drive autonomously on selected roads with suitable conditions, for example without oncoming traffic, cyclists and pedestrians.
Making this complex system 99% reliable is not good enough. You need to get much closer to 100% before you can let self-driving cars mix with other road users in real-life traffic. Here, we have a similar approach to that of the aircraft industry. Our fail-operational architecture includes backup systems that will ensure that Autopilot will continue to function safely also if an element of the system were to become disabled.— Erik Coelingh
For example, the probability of a brake system failure is very small, but a self-driving vehicle needs a second independent system to brake the vehicle to a stop, as it is unlikely that the driver will be prepared to press the brake pedal.
On the road, the complete technology solution will handle even the most complicated scenarios, from smooth commuting to heavy traffic and emergency situations.
Just as good drivers would, potentially critical situations are approached with sensible caution. In a real emergency, however, the car reacts faster than most humans.—Erik Coelingh
When autonomous driving is no longer available—due to exceptional weather conditions, technical malfunction or the end of the route has been reached—the driver is prompted by the system to take over again. If the driver is incapacitated for any reason and does not take over in time, the car will bring itself to a safe place to stop.
In addition to simplifying people’s lives and transforming the everyday commute from lost time to quality time, self-driving cars create environmental benefits. Volvo Cars expects that autonomous driving could cut fuel consumption. The technology could also improve traffic flow as well as open up possibilities for urban planning and more cost-efficient investments in infrastructure.
Developing a complete technological solution for self-driving cars is a major step. Once the public pilot is up and running, it will provide us with valuable knowledge about implementing self-driving cars in the traffic environment, and help us explore how they can contribute to sustainable mobility. Our smart vehicles are a key part of the solution, but a broad societal approach is vital to offer sustainable personal mobility in the future. This unique cross-functional co-operation is the key to a successful implementation of self-driving vehicles.—Erik Coelingh
Drive Me system solution components include:
Sensor technologies. Volvo Cars is developing a holistic solution that generates exact positioning and a complete 360° view of the car’s surroundings. This is achieved by a combination of multiple radars, cameras and laser sensors. A redundant network of computers processes the information, generating a real-time map of moving and stationary objects in the environment.
Precise positioning is based on this surround information together with GPS and a high definition 3D digital map that is continuously updated with real-time data. The system is reliable enough to work without requiring driver supervision.
Combined radar and camera. The combined 76 GHz frequency-modulated continuous wave radar and camera placed in the windshield is the same as that in the all-new XC90. This system reads traffic signs and the road’s curvature and can detect objects on the road such as other road users.
Surround radars. Four radars behind the front and rear bumpers (one on each corner of the car) are able to locate objects in all directions. By sweeping both left and right, transmitting waves that bounce off signs, poles, and tunnels, they monitor a full 360° around the car.
360° surround vision. Four cameras monitor objects in close proximity to the vehicle. Two are under the outer rear-view mirrors, one is in the rear bumper and one is in the grille. Besides detecting objects at close range, these cameras monitor lane markings. The cameras have a high dynamic range and can handle very quick changes in lightning conditions, e.g. when entering a tunnel.
Multiple beam laser scanner. This sensor system is placed in the front of the vehicle, below the air intake. The scanner can identify objects in front of the car and ensures very high angle resolution. It can also distinguish between objects. The unique laser sensor has a range of 150 meters for vehicles and covers a 140° field of view.
Trifocal camera. In addition, a trifocal camera placed behind the upper part of the windscreen is three cameras in one, providing a broad 140° view, a 45° view and a long-range, yet narrow, 34° view for improved depth perception and distant-object detection. The camera can spot suddenly appearing pedestrians and other unexpected road hazards.
Long-range radars. Two long-range radars placed in the rear bumper of the car ensure a good rearward field of view. This technology is particularly useful when changing lanes because it can detect fast-moving vehicles approaching from far behind.
Ultrasonic sensors. Twelve ultrasonic sensors around the car are used to identify objects close to the vehicle and support autonomous drive at low speeds. The sensors are based on the technology used for current park assist functions enhanced with advanced signal processing. A typical example of when this technology is useful is for detecting unexpected situations, such as pedestrians or hazards on the road close to the car.
High definition 3D digital map. A high definition 3D digital map is the tool used to provide the vehicle with information about the surroundings, e.g. altitude, road curvature, number of lanes, geometry of tunnels, guard rails, signs, exits, etc. The position geometry is in many cases at centimeter level.
High performance positioning. The high performance GPS is one part of the positioning control that is enhanced by a combination of an advanced GPS, a 3-degrees of freedom accelerometer and a 3-degrees of freedom gyro. By matching the 360˚ image created by the multitude of sensors with the map image, the car will get the information about its position in relation to the surroundings. By combining the information from the sensors and the map, the Drive Me car is able to choose the best course in real time, factoring in variables such as the curvature of the road, speed limit, temporary signs and other traffic.
Cloud services. The cloud service is connected to the traffic authorities’ control centre. This ensures that the most up-to-date traffic information is always available. The control center operators also have the ability to tell the drivers to turn off the autonomous drive mode if necessary.
This is a big and present surprise for me. It is more evidence that self-driving cars are coming and coming sooner than many think. Musk lately said that Tesla could make an autonomous car by 2020 and then need another 2 to 3 years for testing, validation and authorization of the car in cooperation with authorities to get it approved for driving on public roads. So it seems Volvo currently is 3 years ahead of Tesla in this regard. If Volvo can pull this pilot program off and log several million miles from 2017 to 2019 with a minimal number of incidents in a large variety of driving conditions then it seems that Volvo could win the honor of being the first automaker in the world to sell a fully autonomous vehicle in 2019 or 2020 that is approved for such by all the relevant authorities.
Posted by: Account Deleted | 19 February 2015 at 10:37 AM
A hand to Volvo but Tesla, Apple, Google and many others will not be far behind.
Hope that driverless e-cars will be availble by 2020 or shorthly thereafter. It will be part of a major industrial revolution.
Posted by: HarveyD | 19 February 2015 at 12:39 PM
Previously autonomous vehicles might do alright in clear conditions, but did not have a clue what was happening if the weather closed in.
Now Sweden in the winter is not San Diego, so Volvo obviously feel they have a handle on this, but I note their comment about exceptional weather stopping the system.
Posted by: Davemart | 19 February 2015 at 01:02 PM
Coelingh> "Making this complex system 99% reliable is not good enough. You need to get much closer to 100% before you can let self-driving cars mix with other road users in real-life traffic. "
It's one thing to implement lane-keeping on the freeway. Entirely another to pre-program responses to the thousands of various edge cases that do not occur frequently, but could be deadly when they do. This isn't going to be a Roomba running around your living room.
There will be a large chasm to cross before this becomes viable for city streets.
"if you want to create an impact in the real world, you have to design and produce a complete system that will be safe, robust and affordable for ordinary customers."
Posted by: electric-car-insider.com | 19 February 2015 at 09:07 PM
The autopilot in Tesla's Model S includes one forward radar, 12 long range ultrasonic sensors, one forward looking camera.
For comparison Volvo's autonomous car includes, 7 radars, 12 ultrasonic sensors, 6 cameras and one laser scanner. Moreover, Volvo's autonomous system is redundant so if it breaks a backup system takes over. Tesla's autopilot is not redundant. If it breaks the human driver must take control or the car crashes.
The only point where Volvo is at least 3 years behind Tesla is that Volvo does not have a BEV on the market. Autonomous systems makes sense for all types of cars but mostly for BEVs because it solves once and for all the problem with limited range and long charging time. You simply change vehicle in seconds to move beyond the range of the first autonomous BEV that transported you. Problem solved.
Posted by: Account Deleted | 19 February 2015 at 11:07 PM
Just read this on Bloomberg. Apparently Apple want to launch a BEV in 2020 and currently has 200 people working on it with a budget to hire up to 1000. Apple is poaching employees from all the auto-motive battery makers and people who are experts in robots for auto manufacturing and also Tesla.
I think it is now safe to say that Apple definitely want to make a BEV by 2020. I think Apple could launch a BEV in that time frame with an advanced autopilot system. However, I doubt the "iTransporter 1" will be fully autonomous or sell in big numbers (there is simply not enough time for that not even with Apple's resources). Volume production of Apple BEVs with autonomous driving will come a few years later by iTransporter 2 or iTransporter 3 that will not require a driver license to buy or to use as a taxi service.
Posted by: Account Deleted | 19 February 2015 at 11:49 PM
Below are the calculations for the business case for an autonomous taxi BEV. Furthermore, the cost to drive one mile in an autonomous BEV taxi is compared to the cost of owning a Toyota Camry and the cost of using a Camry with a human taxi driver.
The conclusion is that the autonomous BEV taxi will cost you 0.16 USD per mile to drive. The self-owned Camry will cost you 0.41 USD per mile or about 500 USD per month if you drive 15,000 miles per year (= ($0.41*15,000/12). Finally, the human operated taxi Camry will cost you 1.41 USD per mile which is representative of actual taxi rates.
Reducing the cost of taxi driving from 1.41 USD to 0.16 USD per mile using an autonomous BEV taxi is simply revolutionary. The world will change for the better as a result. Fewer traffic accidents, no air pollution from land transportation, no import of oil for making transportation fuels, no wasted time by traffic congestion. Time spend for transportation can be used productively to sleep, eat, work or for entertainment or education. The average American household can reduce transportation expenses from 500 USD per month per car needed in household to 200 USD per car (= ($0.16*15,000/12). This is as big as it gets for the automotive industry.
Documentation for costs to drive one mile:
1) Life cost to own Toyota Camry: 65,133 USD = (23,000 USD for Camry + 16,000 USD for life gasoline + 4,800 USD for life maintenance + 21,333 USD for life car insurance).
Life cost per mile: 0.41 USD = $65,133/160,000 miles service life.
2) Life cost of Toyota Camry with human taxi driver: 225,133 USD = (23,000 USD for Camry + 16,000 USD for life gasoline + 4,800 USD for life maintenance + 21,333 USD for life car insurance + 160,000 USD for taxi driver).
Life cost per mile: 1.41 USD = $225,133 /160,000 miles service life.
Now consider a fully autonomous taxi with an ultra durable 24kwh lithium titanate battery (10,000 cycles) giving it about 85 miles of range and a service life of 850,000 miles. With autonomous driving the range issue and charging time issue no longer exists as you can change the vehicle in seconds to go an additional 85 miles and keep doing it until you reach your destination. This is the BEV conception that will wipe out any gasser on the market because its total cost per mile is unbeatable by any gasser.
3) Life cost of autonomous BEV taxi: 133,467 USD = (35,000 USD for BEV taxi + 28,800 USD for life electricity + 17,000 USD for life maintenance + 56,667 USD for life car insurance - 4000 USD scrap value of battery).
Life cost per mile: 0.16 USD = $133,467/850,000 miles service life.
Add 1) Toyota Camry assumptions: 1) Service life is 160,000 miles. 2) Long-term price of gasoline is 3 USD. 3) It gets 30 mpg so 16,000 USD spend on gasoline = (160,000/30)*$3. 4) Maintenance cost for oil change, tires, brakes, coolant, etc is 300 USD per 10,000 miles so 4,800 USD = (160,000/10,000)*300 USD. 5) Insurance cost is 2000 USD per 15000 miles so life car insurance is 21,333 USD = (160,000/15,000)*$2000.
Add 2) Toyota Camry taxi assumptions: 1) Service life is 160,000 miles. 2) Long-term price of gasoline is 3 USD. 3) It gets 30 mpg so 16,000 USD spend on gasoline = (160,000/30)*$3. 4) Maintenance cost for oil change, tires, brakes, coolant, etc is 300 USD per 10,000 miles so 4,800 USD = (160,000/10,000)*300 USD. 5) Insurance cost is 2000 USD per 15000 miles so life car insurance is 21,333 USD = (160,000/15,000)*$2000. 6) Hourly pay to chauffeur is 20 USD and hourly markup for time wasted and taxi company overhead is another 20 USD. Operating hours in service for life of car assuming 40 mph is 4000 hours =(160,000/40) so total life cost of chauffeur and taxi company overhead is 160,000 USD = (4000*($20+$20)).
Add 3) Fully autonomous BEV taxi assumptions: 1) Service life is 850,000 miles (= 85 miles battery range*10,000 deep cycles) which corresponds favorably to warranty for Toshibas lithium titanate batteries (see http://www.scib.jp/en/product/detail.htm). 2) 0.28kwh is used to drive one mile (=24kwh battery/85miles range), 3) electricity cost is 28,800 USD = (12 cents per kwh * 0.28kwh* 850,000 miles) which could be much lower off peak, 4) maintenance cost for tires, brakes, coolant, etc is 200 USD per 10,000 miles so 17,000 USD = (850,000/10,000)*200 USD, 5) scrap value of battery after 850,000 miles is 4000 USD. 6) Insurance cost is 1000 USD per 15000 miles so life car insurance is 56,667 USD = (850,000/15,000)*$1000. The lower car insurance for autonomous vehicles assumes that they are twice as good as human drivers to avoid accidents. 7) The Leaf sized vehicle costs 35,000 USD with a 24kwh battery. It is assumed 12000 USD (=24kwh*$500) can be attributed to the battery pack, 6,000 USD for autonomous technology (computers, sensors and redundancy of critical systems) and 17,000 USD for other car expenses. All costs are including gross margins.
Posted by: Account Deleted | 20 February 2015 at 12:20 AM
Sorry for this posting spree. However, I just watched the 30 minute video where Volvo presents their autonomous driving system and it is really informative. I can highly recommend seeing it.
Apparently, Volvo's ambition is not to build an autonomous system for all driving conditions. Sophisticated parking scenarios and city driving is not included. They focus on making an autonomous highway commute possible and making a system that makes crashing a Volvo with deadly outcome nearly impossible wherever you drive. Volvo's time frame is to start selling such cars is 2020. The first Volvo to get this system will of cause be the new Volvo XC90 SUV PHEV which is also the one that is used in the pilot program with the 100 cars in Sweden.
Posted by: Account Deleted | 20 February 2015 at 01:26 AM
From my impression of what Tesla's top priority is with regard to the capability of their autopilot system it is to make an autonomous parking system that can park the Model S and Model X on private property (where normal legislation does not prevent such systems) and automatically plug-in the car for charging. The second top priority will be to get all of Tesla's public parking lots with super charger stations approved for autonomous parking and automatic charging by the authorities. This can be accomplished within 36 months globally IMO.
The fully autonomous car will be a stepwise project getting more capabilities as time allows for its development and proper authorization. Expensive luxury cars are obviously getting the most advanced systems before the lower cost cars. I think it may not be until 2025 that we will see truly autonomous cars that can go anywhere (even off road) and that are made without a steering wheel and a gas pedal and that can be owned and used by people without a driver license for instance kids, disabled people and elderly.
Posted by: Account Deleted | 20 February 2015 at 02:19 AM
Thanks for posting the link to the video Henrik.
At about 18:45, Dr Coelingh talks about realistic expectations for the time frame. "it will not grow from 100 vehicles to 100,000 vehicles overnight. It will take quite a long time before that will happen".
They go on to emphasize the limited scope of the capabilities. (only proscribed roads without pedestrians or cyclists).
Interesting and exciting opportunities, but unrealistic expectations of the scope of capabilities and time frames will not help public understanding or acceptance.
Posted by: electric-car-insider.com | 20 February 2015 at 07:40 AM
This is a link to a recent Nissan announcement about autonomous cars developed at Nissan I quote
"The work of NASA and Nissan – with one directed to space and the other directed to earth, is connected by similar challenges," said Carlos Ghosn, president and CEO of Nissan Motor Co. "The partnership will accelerate Nissan's development of safe, secure and reliable autonomous drive technology that we will progressively introduce to consumers beginning in 2016 up to 2020." Nissan has set 2020 as the timeframe for the introduction of autonomous drive vehicles that have the ability to navigate in nearly all situations, including the most complex situation, city driving."
It seem that Nissan is ahead of Volvo because they believe autonomous city driving is also doable in cars to consumers in 2020. That is the same day Tesla believe they can make an autonomous Tesla that can begin road testing and authorization processes with authorities which is expected to take until about 2023.
Volvo may not be able to sell 100,000 limited autonomous XC90 SUV PHEV in 2020 as they are not a big automaker. However, Nissan and Tesla can. Tesla's capacity in 2020 could be 500,000 BEVs and it could be much more in 2023 when Tesla believe their fully autonomous technology is ready for consumers. I expect all of Tesla's cars to come with this technology. I also expect Apple's car to have it by 2023 or perhaps 2025. It takes no time to manufacture the necessary sensors and chips once the software is done. So when the technology is ready it will be on all luxurious cars made by those who has the technology.
Posted by: Account Deleted | 20 February 2015 at 09:49 AM
I appreciate your optimism Henrik, but these pronouncements about fully-autonomous driving remind me of the old John Scully 1987 keynote demo video of "Knowledge Navigator" (portrayed as far more sophisticated than Siri).
Almost 30 years later, and we're not anywhere close. That's not to say what we have isn't good. It just doesn't map to the totally unrealistic expectations set by that rosy view of future tech.
Note the cautious (and quite practical) qualifications made by the Volvo system designers in the video you posted. This is interesting, important work. It will make cars safer. But I don't expect to be able to nod off and still arrive at my destination anywhere close to 2020.
Posted by: electric-car-insider.com | 20 February 2015 at 05:11 PM
I won't belabor this metaphor further than this post, but here's an interesting counterpoint - the "Knowledge Navigator" video that was created as a result of a discussion with engineers, rather than the non-technical CEO as the first one was. Except that very few people ever saw this one:
Posted by: electric-car-insider.com | 20 February 2015 at 05:27 PM
I did a little research to see who says what about autonomous driving. What I found is that Nissan is the leader already selling self-driving cars that can handle highway traffic and these cars will get fully autonomous (that is, handle the most difficult city driving) by 2020 if Nissan can meet its own deadlines. Many other automakers are not far behind perhaps 2 to 5 years and will also have fully autonomous cars on the road before 2025 including Tesla, Audi, Volvo, Mercedes and Ford. My guess is that Google, Apple and Uber will also start selling autonomous BEVs and transportation services from them before 2025. After 2025 there will be massive volume growth for autonomous travel. By 2035 the global auto industry will exclusively make vehicles that are fully autonomous. IMO not all will be BEVs because it probably takes more time than 20 years to build all the factories that make batteries, power electronics and electric motors. However, in 2035 gassers will be a rapidly dying industry even the PHEVs.
As I see it the fully self-driving system is the silver bullet that will make BEVs price competitive with the cheapest gassers and also make BEVs fully useful because it also solves their range issue and the issue with long charging times. Nissan and Tesla knows it. I believe Apple, Uber and Google knows it too.
1) Nissan say they will start selling fully autonomous cars by 2020 that is capable of urban driving. They have said it repeatedly since 2013. Nissan's Infiniti Q50 is already able to drive it selves at the highway and some you tube videos can be found showing people illegally leaving the driver's seat at full highway speed just to prove it. Nissans research and testing is clearly focused on making autonomous city driving possible in all situations as Nissan believe they got autonomous highway driving covered. It will be interesting to see if Nissan can meet their own deadline and be allowed to sell these fully autonomous cars by 2020.
2) Audi say they will sell their first fully autonomous car by 2017 and the system will be available in the new Audi A8. However, the source is not precise about autonomous driving also include city driving. I quote " Moser (Audi Head of Product and Technology Communications) said Audi "wants to be first" to market with a self-driving car, noting that its autonomous car technology already works well, and it was just a matter of waiting for the legislation to catch up with the technology"
3) Elon Musk say Tesla will build a fully autonomous car by 2020 and have it properly tested and authorized for public roads by 2023.
4) Ford's CEO Mark Fields expect the first fully autonomous cars (including city driving) to hit the roads by 2020 but does not promise that it will be a Ford. He is focused on developing a fully autonomous system that will cost so little that it can be used in all of Fords vehicles not just the luxury vehicles. That will take more time to develop because you may have to stick to inexpensive sensors such as cameras and ultrasonic sensors and forget about radars and laser scanners. Humans do not have radars and laser scanners so they are really not needed if you want to make something that can do better than human drivers.
5) Volvo expect to start selling the an autonomous highway capable Volvo XC90 SUV PHEV by 2020. It will also have systems that prevent deadly accidents in city driving but will not be able to drive itself in cities by 2020.
6) Daimler Chairman Dieter Zetsche have lately said fully autonomous vehicles will be in showrooms between 2020 and 2030.
Posted by: Account Deleted | 21 February 2015 at 02:42 AM
Progress is difficult to stop in free societies and Henrik predictions may be correct for future EVs and ADVs (autonomous drive vehicles) give or take a few years.
Posted by: HarveyD | 22 February 2015 at 08:50 AM
The link below is a video from Volvo that describes their pilot project for self-driving cars in less than 3 minutes. I can recommend it. It shows that Volvo's sensor package is far more expanded than the one applied in Model S from Tesla. Hope that Tesla will soon start selling the Model S with an extended sensor package because the one they currently have is not going to be enough to enable truly self-driving features in the future regardless of future software updates for Model S.
Ordinary Volvo customers in the pilot program will be able to do other productive things legally in most driving situations. This s an enormous advantage even though it is not 100% self-driving in all situations, especially city driving. The greatest thing IMO about Volvo's nearly fully autonomous car is that the car can park itself safely if the driver is not responding for whatever reason, for example, is sleeping or is unconscious from a heart stroke. In other words, there can be situations where the car judges it is best to let the human driver take control but if he or she does not the car will simply park it-self. Indeed, Volvo is the first auto company in the world to test such a high degree of self-driving car among ordinary people. The 100 people in the pilot program are not developers at Volvo. They are just ordinary Volvo customers.
Posted by: Account Deleted | 23 February 2015 at 12:53 AM
Another thing to cheer about Volvo's autonomous car is that Volvo has managed to hide all the sensors so that they do not compromise the good look of the car. Take a look a Nissan's or Google's self-driving cars and they still drive around with monstrous looking sensor racks. If that is still necessary you know there is still a lot of work left.
To me it seems that Volvo can make a future 100% self-driving Volvo XC90 SUV PHEV simply by using its current sensor package and by upgrading its software as the software developers iron out the needed code. Volvo customers should be able to upgrade their car's software using their car's build in 4G internet connection just like Model S owners are doing today.
Posted by: Account Deleted | 23 February 2015 at 01:30 AM