Autonomous Vehicles, LEAF and others...

My Nissan Leaf Forum

Help Support My Nissan Leaf Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Via GCC:
Hyundai NEXO fuel-cell vehicles self-drive (Level 4) 118 miles from Seoul to Pyeongchang
http://www.greencarcongress.com/2018/02/20180204-nexo.html

. . . Five Hyundai vehicles completed the journey. Three vehicles are based on Hyundai’s next-generation fuel cell electric SUV NEXO, scheduled to be released in Korea next month, and the other two are Genesis G80 autonomous vehicles. All vehicles are equipped with level 4 self-driving technology, as defined by the SAE international standards, and 5G network technology. . . .

The Hyundai test marks the first time autonomous vehicles have operated on public highways at 110 km/h (68 mph), the maximum speed allowed by law on Korean highways.

The demonstration took place in Seoul on 2 Feb, with the ‘CRUISE’ and ‘SET’ buttons being pressed on the autonomous-driving steering wheel of each vehicle, at which point the cars immediately switched to self-driving mode and began the 118-mile journey to Pyeongchang.

Entering the highway, the vehicles moved in response to the natural flow of traffic, executed lane changes, overtaking maneuvers and navigated toll gates using Hi-pass, South-Korea’s wireless expressway payment system. . . .

During autonomous driving, a high volume of data is processed by the vehicles on board systems, necessitating large power consumption. A fuel cell electric vehicle is able to produce electricity to meet this power consumption, as well as powering the vehicles drive systems. Hyundai says that the fuel cell vehicle is optimal for this type of test. . . .
As mentioned upthread, I've been increasingly seeing the power requirements of autonomy being a possibly significant obstacle to combining them with BEVs. Presumably power needs will decrease over time, so we'll see if that happens in time to avoid this issue.
 
GRA said:
As mentioned upthread, I've been increasingly seeing the power requirements of autonomy being a possibly significant obstacle to combining them with BEVs.
You must be joking. The power capabilities of BEVs are significantly higher than that of H2 FCVs.

Likely you are talking about the impact of the energy consumption of the electronics on range. I find it interesting that you are still coming here in 2018 bellyachng about this. Today, we have the following BEVs available for sale in the market:

- Tesla offers three different cars with over 300 miles of EPA range.
- GM offers a car with over 200 miles of EPA range.
- Nissan offers the LEAF with 150 miles of EPA range.
- Proterra makes a BEV city bus with a 660-kWh battery that has demonstrated the ability to drive 1100 miles on a single charge in ideal conditions. That bus should be able to easily travel over 300 miles in even very challenging conditions.

We also have the following BEVs being prepared for sale in the marketplace soon:

- Tesla is developing two Class 8 Semi trucks, one with nearly 1 MWh of energy storage and 500 miles of range claimed.
- Tesla has announce a new version of the Roadster which should have over 600 miles of EPA range.

Seriously, let it go. The days of BEVs being limited to 60 miles of energy storage are quickly moving into history. And they do it without wasting over half of the energy provided to them.
GRA said:
Presumably power needs will decrease over time, so we'll see if that happens in time to avoid this issue.
What issue? Of course computing power requirements decrease with time. That's a given. And I posted about massive improvements to the cost (and I will assume also the power consumption) of LiDAR in the post immediately preceding yours.

When you see that something consumes a lot of electrical energy, you would be much better off worrying about how that energy will be efficiently sourced with a minimum of impact to the environment. H2 FCVs do it with the MOST impact of all the options available today. BEVs do it with the least impact and they are improving very rapidly.

In any case, as long as autonomy consumes massive amounts of power, that means it is extremely expensive and will either remain in the lab or will be confined to very few niche applications. it will need to be productized before it will see widespread application.

I do find it interesting that Hyundai is already demonstrating Level 4 Autonomy. Let's not get confused because Hyundai points out a corner case where H2 FCVs have a very slight, temporary advantage over BEVs in one insignificant, also temporary area and entirely lose sight of the massive drawbacks that technology brings to the table.
 
A self-driving truck just drove from Los Angeles to Jacksonville:
https://www.cnbc.com/2018/02/06/embark-trucks-self-driving-truck-drives-los-angeles-to-jacksonville.html

The company involved is Embark, which I'd not heard of, until now.
 
RegGuheert said:
GRA said:
As mentioned upthread, I've been increasingly seeing the power requirements of autonomy being a possibly significant obstacle to combining them with BEVs.
You must be joking. The power capabilities of BEVs are significantly higher than that of H2 FCVs.

RegGuheert said:
Likely you are talking about the impact of the energy consumption of the electronics on range.
Yup. Should have written energy.

I find it interesting that you are still coming here in 2018 bellyachng about this.
Uh, still? I hadn't even considered it as an issue until I read "Driverless: Intelligent Cars and the Road Ahead" last year, and then started to see it mentioned more and more with respect to BEVs, with some calcs showing the effects on range. I have no opinion either way, as I don't know enough about the current state of the art.

RegGuheert said:
Today, we have the following BEVs available for sale in the market:

- Tesla offers three different cars with over 300 miles of EPA range.
- GM offers a car with over 200 miles of EPA range.
- Nissan offers the LEAF with 150 miles of EPA range.
- Proterra makes a BEV city bus with a 660-kWh battery that has demonstrated the ability to drive 1100 miles on a single charge in ideal conditions. That bus should be able to easily travel over 300 miles in even very challenging conditions.

We also have the following BEVs being prepared for sale in the marketplace soon:

- Tesla is developing two Class 8 Semi trucks, one with nearly 1 MWh of energy storage and 500 miles of range claimed.
- Tesla has announce a new version of the Roadster which should have over 600 miles of EPA range.

Seriously, let it go. The days of BEVs being limited to 60 miles of energy storage are quickly moving into history. And they do it without wasting over half of the energy provided to them.
No one's talking about 60 mile BEVs, they're talking about longer range BEVs (and other cars), using significant amounts of battery energy to run the sensors and computers. If BEVs need to use a large proportion of their electrical energy to run the autonomous systems, they are obviously the tech whose range will be most affected. Whether or not that range reduction will be judged serious or not by consumers and companies remains to be seen.

RegGuheert said:
GRA said:
Presumably power needs will decrease over time, so we'll see if that happens in time to avoid this issue.
What issue? Of course computing power requirements decrease with time. That's a given. And I posted about massive improvements to the cost (and I will assume also the power consumption) of LiDAR in the post immediately preceding yours.
And the hope is that all the other sensors will also decrease, so that this will prove a minor issue. We aren't there yet, so we don't know how the timing will go.

RegGuheert said:
When you see that something consumes a lot of electrical energy, you would be much better off worrying about how that energy will be efficiently sourced with a minimum of impact to the environment. H2 FCVs do it with the MOST impact of all the options available today. BEVs do it with the least impact and they are improving very rapidly.

In any case, as long as autonomy consumes massive amounts of power, that means it is extremely expensive and will either remain in the lab or will be confined to very few niche applications. it will need to be productized before it will see widespread application.
Right, so the question is will the timing of autonomy and energy reduction be out of sync or not?

RegGuheert said:
I do find it interesting that Hyundai is already demonstrating Level 4 Autonomy. Let's not get confused because Hyundai points out a corner case where H2 FCVs have a very slight, temporary advantage over BEVs in one insignificant, also temporary area and entirely lose sight of the massive drawbacks that technology brings to the table.
As is usually the case with consumer products, the public will decide which benefits and disadvantages are most important to them. I have no idea which will succeed, and whether or not this will prove a significant handicap for autonomous BEVs (or FCEVs FTM) compared to ICEs. But I posted the info here because of the L4, not because it was an FCEV.
 
GRA said:
But I posted the info here because of the L4, not because it was an FCEV.
You could've fooled me, since the only thing you commented on was how you had to have a fuel cell to do that drive and said NOTHING about the Level 4 capability. And you continue on, as if the issue is a big one.

All Teslas and the 2018 Nissan LEAF have autonomous features availavable for sale TODAY. I wonder how they manage to power all the sensors that they include. GM is targeting the using the Chevy Bolt for all their autonomous vehicle work. To my knowledge, NO H2 FCV available today has these features.

Yes, the sensors in autonous vehicles can and are powered by production BEVs. No, these features are not available in H2 FCVs today.
 
RegGuheert said:
GRA said:
But I posted the info here because of the L4, not because it was an FCEV.
You could've fooled me, since the only thing you commented on was how you had to have a fuel cell to do that drive and said NOTHING about the Level 4 capability. And you continue on, as if the issue is a big one.
I t seems I did fool you. I went back and forth for a while about whether to post it in this topic, the Nexo one or cross-post in both, but ultimately decided that the type of propulsion was less important than the claim of L4 capability, as they put it in both the FCEV and the other vehicles.

RegGuheert said:
All Teslas and the 2018 Nissan LEAF have autonomous features availavable for sale TODAY. I wonder how they manage to power all the sensors that they include. GM is targeting the using the Chevy Bolt for all their autonomous vehicle work. To my knowledge, NO H2 FCV available today has these features.

Yes, the sensors in autonomous vehicles can and are powered by production BEVs. No, these features are not available in H2 FCVs today.
As none of the Teslas, Nissans or Chevys you mention are currently available with autonomy (which is defined by the SAE as L4 or L5) yet, nor is any other vehicle that can be sold to the public and used on a public road without a safety monitor present, I don't agree with your statement. Quite frankly, until I read some of the concerns and details of the energy requirements for full autonomy in "Driverless", it'd never even occurred to me that there might be an issue. I lack the knowledge to say if it will be one when the time comes, but people who know far more about the subject than I do have been bringing it up so I feel it deserves to be mentioned, and have posted links to such. "Driverless" described it as a potential problem (among several others), one which might need to be solved to allow autonomous BEVs to achieve their goals, but the book is about autonomy, not AFVs.
 
Here's a recent article that supports your assertion, GRA:
AutoWeek said:
How much power are we talking about?

Bloomberg says that current prototypes for fully autonomous driving systems consume the equivalent energy of 50 to 100 laptops, citing supplier BorgWarner. This translates to 2 to 4 kilowatts of electricity, which in a modern car makes it 5 to 10 percent more difficult to meet fuel economy and carbon emission targets.
Simply put, that IS a lot of power.

So lets run the numbers and find out whether the assertion that this will keep fully-autonomous BEVs off the road for the near future is a credible one.

First of all, a 4 kW load connected to the 12-V power in a car would draw OVER 300 A of current (assuming a float voltage of about 13 V like we see in the LEAF). I am unaware of ANY vehicle with this much current available at 12 V. That's over twice the current that the DC-DC converter in my 2011 LEAF can put out. Even the local rednecks who "amp up" the electrical systems in their pickup trucks so that they can operate the hydraulics for a snowplow are limited to about 300 A TOTAL.

That explains why these developments are being done almost exclusively on vehicles with electric drivetrains, since those have 10s or 100s of kW of electrical power available from the traction battery.

Even though that power level is for prototypes and such a power-hungry system would NEVER be put into a production car, let's just run the numbers anyway to see what it really would mean for the range of a modern BEV. I will define a 10% hit on range as the limit of what is reasonable. I will also use a 2018 Chevy Bolt as the reference vehicle, as it is what GM is targeting for fully-autonomous operation in 2019. The Bolt has over 60 kWh available but for this calculation, let's use 60 kWh.

10% of 60 kWh is 6 kWh. In other words, you could take advantage of autonomy for 1.5 hours at 4 kW or 3 hours at 2 kW and still have 90% of your range left. Put another way, the range mentioned in the article could easily be met by the Chevy Bolt equipped with an autonomy system that draws 2 kW, but with a 4-kW system it would exceed the 10% limit I set (but likely would still be able to make the drive).

But no automaker would field an autonomous system which draws more power than the resistive heater in my LEAF consumes in 10 F temperatures. That's the difference between a prototype and a product. So, how much power WILL the first fully-autonomous systems draw. I will put the ABSOLUTE upper limit at 1 kW, but I seriously doubt any automaker would ever commit to a system that consumed more than 500 W. Why? Because there are far too many knock-on effects to try to add another kW load into the car. But let's look at the effect of a 1-kW autonomy system on various vehicles' range.

Here are the number of hours that could be driven using a 1-kW autonomy system before 10% of the battery capacity is consumed:

- 2018 Chevy Bolt (60 kWh): 6 hours
- 2018 Nissan LEAF (36 kWh): 3.6 hours
- 2018 Tesla Model 3 (75 kWh): 7.5 hours
- 2018 Tesla Model X P100D (100 kWh): 10 hours
- 2018 Proterra Catalyst E2 max (660 kWh): 66 hours
- 2019 Nissan LEAF (55 kWh): 5.5 hours
- 2019 Tesla Truck (900 kWh): 90 hours
- 2020 Tesla Roadster 2 (200 kWh): 20 hours

Sorry, GRA, but this is not an issue for actual BEVs. Let's not imagine that autonomy is some sort of justification to field H2 FCVs. It is not.
 
First off, let me apologize for stating that I first read about this issue in "Driverless". I thought I had, but re-read the book last night and it wasn't mentioned there, so it must have been somewhere else; sorry for the bad steer. I'm pretty sure I posted two or three links referencing this issue on MNL, either upthread or elsewhere, and will try to find them.

"Driverless" is still well worth the read to anyone interested in AVs, as it discusses the history, technical (AI, sensors), regulatory, legal (who's at fault), societal (jobs, housing) and ethical (the Trolley problem) issues involved in bringing this tech to market, and potential positive and negative effects from same. One of the authors was part of the team from Carnegie Mellon Univ. that took part in the DARPA challenges, and the notes almost all have URLs to the papers etc. cited, so it's a great source of further info. The info on how AI has developed and the difference between symbolic (rules-based) and machine learning AI (esp. deep learning) approaches is particularly useful to anyone who, like me, is unfamiliar with this stuff.

RegGuheert said:
Here's a recent article that supports your assertion, GRA:
AutoWeek said:
How much power are we talking about?

Bloomberg says that current prototypes for fully autonomous driving systems consume the equivalent energy of 50 to 100 laptops, citing supplier BorgWarner. This translates to 2 to 4 kilowatts of electricity, which in a modern car makes it 5 to 10 percent more difficult to meet fuel economy and carbon emission targets.
Simply put, that IS a lot of power.

So lets run the numbers and find out whether the assertion that this will keep fully-autonomous BEVs off the road for the near future is a credible one.

First of all, a 4 kW load connected to the 12-V power in a car would draw OVER 300 A of current (assuming a float voltage of about 13 V like we see in the LEAF). I am unaware of ANY vehicle with this much current available at 12 V. That's over twice the current that the DC-DC converter in my 2011 LEAF can put out. Even the local rednecks who "amp up" the electrical systems in their pickup trucks so that they can operate the hydraulics for a snowplow are limited to about 300 A TOTAL.

That explains why these developments are being done almost exclusively on vehicles with electric drivetrains, since those have 10s or 100s of kW of electrical power available from the traction battery.

Even though that power level is for prototypes and such a power-hungry system would NEVER be put into a production car, let's just run the numbers anyway to see what it really would mean for the range of a modern BEV. I will define a 10% hit on range as the limit of what is reasonable. I will also use a 2018 Chevy Bolt as the reference vehicle, as it is what GM is targeting for fully-autonomous operation in 2019. The Bolt has over 60 kWh available but for this calculation, let's use 60 kWh.

10% of 60 kWh is 6 kWh. In other words, you could take advantage of autonomy for 1.5 hours at 4 kW or 3 hours at 2 kW and still have 90% of your range left. Put another way, the range mentioned in the article could easily be met by the Chevy Bolt equipped with an autonomy system that draws 2 kW, but with a 4-kW system it would exceed the 10% limit I set (but likely would still be able to make the drive).

But no automaker would field an autonomous system which draws more power than the resistive heater in my LEAF consumes in 10 F temperatures. That's the difference between a prototype and a product. So, how much power WILL the first fully-autonomous systems draw. I will put the ABSOLUTE upper limit at 1 kW, but I seriously doubt any automaker would ever commit to a system that consumed more than 500 W. Why? Because there are far too many knock-on effects to try to add another kW load into the car. But let's look at the effect of a 1-kW autonomy system on various vehicles' range.

Here are the number of hours that could be driven using a 1-kW autonomy system before 10% of the battery capacity is consumed:

- 2018 Chevy Bolt (60 kWh): 6 hours
- 2018 Nissan LEAF (36 kWh): 3.6 hours
- 2018 Tesla Model 3 (75 kWh): 7.5 hours
- 2018 Tesla Model X P100D (100 kWh): 10 hours
- 2018 Proterra Catalyst E2 max (660 kWh): 66 hours
- 2019 Nissan LEAF (55 kWh): 5.5 hours
- 2019 Tesla Truck (900 kWh): 90 hours
- 2020 Tesla Roadster 2 (200 kWh): 20 hours

Sorry, GRA, but this is not an issue for actual BEVs. Let's not imagine that autonomy is some sort of justification to field H2 FCVs. It is not.
Thanks for the above, and BTW, it's not my assertion, it's the assertion of articles like that one and the experts they've consulted. As to it being some sort of justification for FCEVs, that remains to be seen. I've mentioned that it might be an issue that could noticeably reduce the range of any AV based on a BEV, which might make the vehicle less acceptable to consumers, and thus shift interest in ZEV AVs to FCEVs. It does look like short-range, low cost privately-owned urban BEVs would be most impacted, especially when the knock-on effects are added (extra weight of the computers, sensors and data storage etc. with the necessary redundancy, structural weight to carry it and any motor power increases to maintain performance), but as long as battery costs come down soon enough it may not matter.

As to how much juice prototypes are using, one of the things that "Driverless" makes clear is the enormous change from L3 or below to L4 and above in terms of computational and sensor power and data storage. These are improving rapidly (companies like Nvidia who make computer game processors leading the way), so the problem may be solved before it becomes an issue. All the recent articles I could find with a Google seem to cite or are at least based on the info in that same Bloomberg article, but arrive at slightly different conclusions. Here's C&D's: https://blog.caranddriver.com/self-driving-vehicles-may-save-energy-despite-power-hog-tech-on-board/

Here's a more recent (4 days ago) article from Wired: https://www.wired.com/story/self-driving-cars-power-consumption-nvidia-chip/

Here's a GCC article:
Continental and NVIDIA partnering on autonomous vehicle systems; market introduction in 2021
http://www.greencarcongress.com/2018/02/20180206-conti.html

NVIDIA and Continental are partnering to create AI (artificial intelligence) self-driving vehicle systems built on the NVIDIA DRIVE platform, with a planned market introduction starting in 2021 for Level 3 functionality. . . .
So, I'd say I'm now leaning cautiously optimistic that it won't be a major issue by the time L4 AVs are actually for sale to the public, but it bears watching.
 
GRA said:
Thanks. Good article.

In the processing world, you can nearly always trade hardware against speed and power issues. But new chip designs do not come cheap, so this is typically only done for specific applications if the addressable market is extremely large. This is just that sort of market.

Eventually these systems will drop well below 100W.
 
Looking at this further, the Xavier AI supercomputer looks a lot like the follow-on to the Tegra X1 processor that Tesla uses in their cars. Here are the main differences:

1) Move from 20 nm to 16 nm process.
2) 512-core Volta GPU rather than 256-core Maxwell GPU
3) 8-core CPU versus 4-core CPU
4) 8K versus 4K video processor

What appears to be new are the "computer vision accelerator" and the "deep learning accelerator".

Here is an excellent primer on what "deep learning" actually is and what it means:

[youtube]http://www.youtube.com/watch?v=ht6fLrar91U[/youtube]
 
Via GCC:
LCA study finds connected and automated vehicle subsystems could enable net reduction in vehicle energy use and GHGs by up to 9%
http://www.greencarcongress.com/2018/02/20180216-lca.html

Researchers at the University of Michigan and Ford Motor Company have found that using a Level 4 connected and automated vehicle (CAV) subsystem could increase vehicle primary energy use and GHG emissions by 3–20% due to increases in power consumption, weight, drag, and data transmission.

However, when potential operational effects of CAVs are included (such as eco-driving, platooning, and intersection connectivity), the net result is up to a 9% reduction in energy and GHG emissions in the base case. Their study is published in the ACS journal Environmental Science & Technology.

  • The major factors that will determine CAV energy outcomes include the following: eco-driving, platooning, lightweighting, and rightsizing, reduced driving to locate parking, ride sharing, congestion mitigation, higher travel demand due to reduced travel cost, increased travel by underserved populations, and faster highway speeds. When these factors are combined in a travel demand and efficiency framework with a 100% CAV adoption scenario, the resulting impact ranges from a 60% reduction in energy consumption up to a 200% increase. Integrating these results with EIA reference case projections for the transportation sector produces light duty vehicle energy consumption estimates between 7,000 and 18,000 trillion Btu in 2050 compared to 16,000 trillion Btu in 2016.

    The existing literature on CAV energy implications … focuses primarily on the operational impact. However, little is known about the full life cycle implications of CAV deployment. Previous studies have omitted the potential vehicle production life cycle impacts with the rationale that they are smaller in magnitude when compared to the travel related energy consumption, but no work has been done to date to examine this assumption.

    To fill the gap, this study provides a life cycle assessment (LCA) of CAV sensing and computing subsystems integrated into both internal combustion engine vehicle (ICEV) and battery electric vehicle (BEV) platforms. The LCA quantifies the burdens from the production and use of Level 4 CAV subsystems applied to the vehicle platforms.


    —Gawron et al. . . .

The team modeled its BEV platform ias the 2015 Ford Focus Electric and the ICEV platform as the 2015 Ford Focus. The small, medium, and large CAV subsystems were based on the configurations in use on the Tesla Model S, Ford Fusion (AV test vehicles), and Waymo’s Chrysler Pacifica, respectively.

The life cycle phases included in the study were materials production, manufacturing and assembly, use, and end of life management. Four factors needed to be considered for the use-phase analysis, the team noted:

  • The increased vehicle energy consumption due to the added electricity demand.

    The increased fuel consumption due to the weight of the CAV subsystem.

    Increased aerodynamic drag due to the use of exterior mounted components.

    The burdens associated with map data transmission over wireless networks. . . .
 
Via GCC:
Autonomous vehicle testing beginning at Marine Corps Air Station Miramar
http://www.greencarcongress.com/2018/02/20180223-miramar.html

Federal, state and local government stakeholders along with industry representatives met to discuss regional and Marine Corps autonomous vehicle proving grounds, to watch autonomous vehicles begin testing, and to tour the installation at Marine Corps Air Station (MCAS) Miramar (located near San Diego) earlier this week.

The representatives came from the San Diego Association of Governments, California Department of Motor Vehicles, California Department of Transportation, California Energy Commission, California Highway Patrol, Space and Naval Warfare Systems Center Pacific (SSCPAC), Qualcomm Technologies, Inc., and MCAS Miramar. . . .

Two routes on the air station, a “rural” and “urban” route, will provide opportunities for stakeholders to initiate autonomous vehicle testing and development projects. The rural route is located on East Miramar where autonomous vehicle technology testing began early February 2018.

The second project, born of a partnership between the Marine Corps and the state, is planned for later this year. It will pilot the use of autonomous vehicles including electric, autonomous vehicles that are also vehicle-to-grid (V2G) capable. This program will assess the feasibility of autonomous vehicles as a means of transportation for official and personal use on the air station and the effectiveness of V2G technology.

Two autonomous, electric 15-passenger vans will follow the preprogrammed, “urban” route, which will cover the installation’s most frequented locations in the morning, afternoon and evening. When not operating on the preprogrammed schedule, the two vans will be available on-demand along the route.

In addition to testing autonomous driving technology, the vans will be V2G capable, which means they can connect to the air station’s microgrid via a bi-directional electric-vehicle charging station. . . .
 
Apparently Hyundai's "level 4" autonomous Nexo isn't:
Autoweek said:
An autonomous prototype of Hyundai's fuel cell powered Nexo crossover was quickly approaching a bustling pedestrian crossing near an entrance to the Winter Olympics here on Feb. 21. Hyundai was demonstrating its Level 4 autonomous vehicle technology, which it plans to commercialize in certain cities by 2021.

I was in the back seat, anticipating how the robotic Nexo would react. The system had been driving conservatively up to that point of the brief demo, so I figured the Nexo would recognize the pedestrians who were being picked up by its sensors and decelerate to a gradual stop.

But that's not what happened.

The Nexo continued forward at deliberate pace while yellow lights flickered above the crosswalk. It showed no signs of stopping.

Then, with the crosswalk just 10 or so feet ahead of us, the engineer in the driver's seat hit the brakes himself to avert potential calamity.
 
RegGuheert said:
Apparently Hyundai's "level 4" autonomous Nexo isn't:
Autoweek said:
An autonomous prototype of Hyundai's fuel cell powered Nexo crossover was quickly approaching a bustling pedestrian crossing near an entrance to the Winter Olympics here on Feb. 21. Hyundai was demonstrating its Level 4 autonomous vehicle technology, which it plans to commercialize in certain cities by 2021.

I was in the back seat, anticipating how the robotic Nexo would react. The system had been driving conservatively up to that point of the brief demo, so I figured the Nexo would recognize the pedestrians who were being picked up by its sensors and decelerate to a gradual stop.

But that's not what happened.

The Nexo continued forward at deliberate pace while yellow lights flickered above the crosswalk. It showed no signs of stopping.

Then, with the crosswalk just 10 or so feet ahead of us, the engineer in the driver's seat hit the brakes himself to avert potential calamity.
Not yet, anyway, but does seem to be getting there. Handling the roundabout by itself is pretty impressive - many Americans have trouble with that, owing to unfamiliarity.
 
Quite modest scale initial AV taxi test by Nissan:

Take that, Uber: Nissan starts live tests of its autonomous taxi

...Today in Yokohama, Nissan started a pilot with self-driving Leaf cars used as autonomous taxis. Partnering with e-commerce company DeNA, Nissan will start its “Easy Ride” service in Yokohama on March 5th, which puts it, according to Reuters, “among the first major automakers anywhere to test ride-hailing software developed in-house, using its own fleet of self-driving electric cars.”

The test has simple beginnings. Cars will self-drive a 4.5 kilometer (2.8 miles) route down the main drag of Yokohama’s Minatomirai district, with the city’s scenic Red Brick Warehouse on one end, and Nissan’s HQ conveniently at the other. Devoid of the narrow alleys Japan is known for, it is a “relatively simple route,” admitted Hiroshi Nakajima, head of DeNA’s automotive business unit. Taking place from “March 5th to March 18th, involving 300 people of the general public,” as Nissan’s CEO Hiroto Saikawa said today, the test will be short, and limited in scope...
https://dailykanban.com/2018/02/take-uber-nissan-starts-live-tests-autonomous-taxi/
 
Table added with vehicles with lane centering tech. Feel free to improve.
https://en.wikipedia.org/wiki/Lane_centering#Sample_of_vehicles_with_lane_centering_assist
 
Via sfgate, the San Francisco Chronicle's web portal:
Look, ma, no driver! Cars without humans coming to California soon
https://www.sfchronicle.com/news/ar...s-coming-12704455.php?cmpid=gsa-sfgate-result

Robot cars without drivers could be on California streets by early spring. Long-awaited state regulations for autonomous cars without human drivers may be approved Monday by a legal compliance agency, according to the state Department of Motor Vehicles, which submitted the rules to the Office of Administrative Law on Jan. 11. A thumbs-up would open a 30-day public notice period, during which companies can apply for permits to test the cars. California could begin issuing such permits April 2, according to spokeswoman Jessica Gonzalez.

“I think we’ll definitely see some companies applying for driverless testing permits right away,” she said. “Manufacturers will want to be the first ones.”

California is a hotbed for robot-car development, with 50 companies testing 387 autonomous vehicles. They all are required to have a backup driver ready to take the wheel on public roads. . . .

Companies must jump through several hoops to get an OK for driverless testing, including certifying that the cars are ready. Robot cars must have a communication link to remote operators who can steer them if they encounter problems. Their owners must show a plan to train those operators. Car makers could contract out the remote-operator function to third parties, such as Mountain View’s Phantom Auto, which is developing remote-control centers for robot cars and said it has several interested customers. . . .
 
Uber Halts Autonomous-Car Testing After Fatal Arizona Crash
https://www.bloomberg.com/news/articles/2018-03-19/uber-autonomous-car-involved-in-fatal-crash-in-arizona

Uber halts self-driving tests after pedestrian killed in Arizona
A woman was killed early Monday morning in Tempe
https://www.theverge.com/2018/3/19/17139518/uber-self-driving-car-fatal-crash-tempe-arizona
 
Back
Top