16th edition of the 6th year of SmartDrivingCars
Thursday, April 4, 2018
Waymo Isn’t Going to Slow Down Now
M. Bergen, “Apr 2, ” Waymo, the self-driving car company started by Google, did nothing after an autonomous vehicle run by Uber killed a pedestrian in Tempe, Arizona. It didn’t pull back on tests in the nearby suburb of Chandler, where passengers are already taking rides with no one behind the wheel. Its fleets elsewhere didn’t abandon public streets, a precautionary move made by Toyota. For Krafcik, the crash video validated the philosophy Waymo had been following long before he joined, back when it was still part of Google: Never trust humans in cars….
Waymo is now nearing a final deal with a third automaker, Honda Motor Co., in a move that will test the company’s ability to compete in the $164 billion delivery and logistics market. The delivery focus of the alliance with Honda hasn’t been previously reported. The companies have been silent since announcing talks in late 2016, but results are coming soon. …
For his first initiative, Krafcik tried to cut an ambitious deal with a former employer: Ford. The two companies conceived a plan with new self-driving vehicle designs and a multi-year business alliance. The talks collapsed after Mark Fields, Ford’s then-CEO, flew to Google’s headquarters for a dinner with Page where he pushed for a faster timeline, according to a person familiar with the episode.
Krafcik said Waymo walked away because the terms were unfair. “It just wasn’t the right one for us,” he said. “We already had the risk of the technology. We’re also deciding on vehicle formats for years and years and years. There was significant capital risk on our plate, not shared equally.” A Ford spokesman declined to comment. This was also the moment when Krafick discovered brewing rancor within Waymo. Anthony Levandowski, a founding engineer on the Google self-driving project, had simmering disputes with Chris Urmson, the program’s longtime leader. Krafcik was brought in to “referee a cage match,” a former employee said. Tensions boiled over, and Levandowski left—but not before sending an email to Page that criticized Krafcik and the attempted Ford deal and proposed splintering the car team. …
Krafcik said that future deals with Hyundai, Ford and others are all possibilities, and he has talked to companies in China, too. (He said he hasn’t spoken with Uber CEO Dara Khosrowshahi since the settlement of the lawsuit.) He now sees Detroit, after years of fearing Google, feeling less threatened. “Very much so,” he said.
Some onlookers question if Krafcik will be around to see Waymo’s alliances through. “You can’t meet John,” said Noble, the consultant, “and not think he’s someone that would have fun running a carmaker.”
For now, though, Krafcik looks to be having fun running a company that’s resolutely not making cars. On the convention floor in Las Vegas, he spotted a Ford Transit Wagon. It’s a hulking eight-seat model he worked on years ago that looks best suited for shuttling around a troop of Girl Scouts or a military platoon.
Krafcik leaped into the second row and turned to the nearest Ford employee: “Do you have a self-driving version?” The answer was no. “Coming soon,” Krafcik said with a laugh.” Read more Hmmmm… Wow, this is more info than has been put out by Google/Waymo in the previous 9 years combined. Looks like Waymo has entered the market/sales phase of its metamorphosis. By the way, who gets to benefit from the deployment of the 1st 20k of the Jaguars. Phoenix and Mountain View don’t have enough demand. Is there going to be a competition a la the frenzy created by the “who wants the 2nd Amazon HQ”? Alain
Smart Driving Cars Podcast Episode 33
F. Fishkin, Apr 4, ” Waymo is making it real! In Episode 33 of the Smart Driving Cars Podcast, hosts Fred Fishkin and Princeton’s Alain Kornhauser are joined by Michael Sena, publisher of The Dispatcher newsletter. Take a deep dive into Waymo’s deals with Jaguar and talks with Honda.. Tesla, Volvo, Uber and Ambarella. And the Princeton Smart Driving Car Summit is coming up! ”
Hmmmm…. Now you can just say “Alexa, play the Smart Driving Cars podcast!” . Ditto with Siri, and GooglePlay. Alain
Real information every week. Lively discussions with the people who are shaping the future of SmartDrivingCars. Want to become a sustaining sponsor and help us grow the SmartDrivingCars newsletter and podcast? Contact Alain Kornhauser at firstname.lastname@example.org! Alain
2nd Annual Princeton SmartDrivingCar Summit
Application NOW OPEN
Become a Sponsor and Promote your Wares
Waymo and Honda reportedly will build a self-driving delivery vehicle together
A. Hawkins, Apr 2, “Waymo has been negotiating a partnership with Honda since late 2016, when the company was still housed within Google… rather than moving people, …., the partnership with Honda will focus on delivering goods.
Waymo CEO John Krafcik wouldn’t reveal much, but the implication is that a deal should be announced soon. …
Krafcik said not to expect the new service to take the form of a “traditional car driven on roads.” His comments suggest Waymo is ready to try co-creating a vehicle from scratch with an automaker rather than modifying existing models, as it has done with Jaguar and Fiat Chrysler. The Honda model may move people and goods, Krafcik hinted; it might be smaller than a truck and could come without a steering wheel or brakes. A Honda spokesman said the companies are “continuing to explore” the relationship.
An entirely new vehicle built from the ground up in partnership with a major OEM would be new territory for the self-driving company, which traditionally has collaborated with automakers to outfit existing models with its self-driving technology….” Read more Hmmmm… Seems very unlikely that such a development would be done without Amazon. The partnering of these two giants to deliver all I need to my front door in the wee hours of the night by driverless vans using otherwise empty streets, with no children, pedestrians or dogs around seems utterly brilliant. Amazon can’t risk “Uber-mistakes”, nor has the time, to play “Driverless-catchup” and Google/Alphabet/Waymo doesn’t have any packages to move. This is a no-brainer. Alain
Tesla Autopilot confuses markings toward barrier in recreation of fatal Model X crash at exact same location
F. Lambert, Apr 3. “Yesterday, we reported on a Tesla owner almost crashing on video trying to recreate the fatal Autopilot accident that happened in Mountain View last month….. As the owner in Chicago showed yesterday, Autopilot’s Autosteer stayed locked on the left line even though it became the right line of the ramp, which led the car into the divider and toward the barrier.”’
The main difference between the two tests is that the confusion with the lanes happened a longer distance ahead of the barrier than in the similar test in Chicago. In the previous test, it almost led the driver into a crash as he was holding the camera at the same time – something we don’t recommend.
As for the recreation in Mountain View, the driver clearly has more time to correct the Autopilot back into the lane.In its report based on the data logs, Tesla claimed:
“The driver had about five seconds and 150 meters of unobstructed view of the concrete divider with the crushed crash attenuator, but the vehicle logs show that no action was taken.”
It seems about right from the video – though it appears to be more than 5 seconds at the speed that the Model S driver was driving….” Read more Hmmmm… Also See video. What continues to bother me is why the Automated Emergency Braking (AEB) System never applied the brakes. The car is heading straight into a fixed object on what it thinks is and has planned as its road ahead and no brakes????
Also, the lane markings are confusing. They could easily confuse us but we’d look up and see the barrier and apply the brakes. State DoTs: we are asking very little from your infrastructure except good lane markings, readable signs and smooth running surfaces. These are the exact same thing that human drivers have been wanting for years. We don’t want, nor need, V2X gizmos and IoTs, just paint, readable signs and no pot holes. Please!!! Alain
Tesla Autopilot confuses markings toward barrier in recreation of fatal Model X crash at exact same location
F. Lambert, Mar 30, “…What we knew happened according to a report from the California Highway Patrol, a Model X hit the median barrier on the highway 101 in Mountain View and it quickly caught on fire and got hit by two other cars.
After reviewing the logs, Tesla elaborated in a new blog post:
“In the moments before the collision, which occurred at 9:27 a.m. on Friday, March 23rd, Autopilot was engaged with the adaptive cruise control follow-distance set to minimum. The driver had received several visual and one audible hands-on warning earlier in the drive and the driver’s hands were not detected on the wheel for six seconds prior to the collision. The driver had about five seconds and 150 meters of unobstructed view of the concrete divider with the crushed crash attenuator, but the vehicle logs show that no action was taken.”
It’s still unclear how the vehicle moved from the lane to the barrier. In its previous statement about the accident, Tesla claimed that Autopilot successfully completed “over 200 trips per day on this exact stretch of road.” Read more Hmmmm…Again, why didn’t the AEB apply the brakes???? Alain
Tesla defends having to release its Autopilot data logs after being criticized for ‘blaming customers’
F. Lambert, Apr 3, “A misleading article titled ‘The customer is always wrong: Tesla lets out self-driving car data – when it suits‘ in the Guardian today sparked criticism over Tesla’s policy about releasing Autopilot data following a few crashes where the drivers blamed Tesla’s driver assist system. I say that it’s misleading because the author, Sam Thielman, claims that we are talking about “self-driving” vehicles, which is obviously not the case….
What he is … talking about are the logs that Tesla has sometimes released in the media after drivers have claimed that some of the company’s Autopilot features have caused crashes. Thielman asserts that Tesla conveniently only releases the data when it serves them – hence the “customer is always wrong” comment.
…Tesla’s owner manual has a warning especially for the kind of situation that led to the accident:
“Warning: Traffic-Aware Cruise Control can not detect all objects and may not brake/decelerate for stationary vehicles, especially in situations when you are driving over 50 mph (80 km/h) and a vehicle you are following moves out of your driving path and a stationary vehicle or object, bicycle, or pedestrian is in front of you instead. Always pay attention to the road ahead and stay prepared to take immediate corrective action. Depending on Traffic-Aware Cruise Control to avoid a collision can result in serious injury or death. In addition, Traffic-Aware Cruise Control may react to vehicles or objects that either do not exist or are not in the lane of travel, causing Model S to slow down unnecessarily or inappropriately.”…Read more Hmmmm…There are a couple of issues here.
1. Release of pre-crash data: My view is (and it may need legislation/regulation to make it enforceable)… If an automated vehicle is involved in a crash on a public roadway, all of the data that it used for the, say, 4 seconds before the crash should be archived and released to the public so that, if possible, improvements can be made that will reduce the likelihood of the recurrence of the crash. Everyone MUST be able to learn from each crash. Full transparency is absolutely necessary.
2. If the AEB isn’t going to apply the brakes if there is a stationary object on the road ahead, then drivers should be warned not to place their feet very far away from the brake and they MUST constantly pay attention. In fact, if any “Self-driving” function is engaged, the car should NOT be able to accelerate to a speed above which the AEB can reliably detect pedestrians and stationary objects in the lane ahead. AND, those functions MUST not be “engageable” if the car is traveling at or above that threshold speed. Let’s be responsible here. It would only take a couple lines of code. Alain
Tesla owner almost crashes on video trying to recreate fatal Autopilot accident
F. Lambert, Apr 3, “…now another Tesla owner also conducted his own little investigation into the accident by following a similar scenario on Autopilot and almost crashed on video by doing so….We can see the driver ignoring an alert to ‘hold the steering wheel’ sent out a few seconds before the barrier just like Tesla said in its report based on the logs – though that was likely a time-based alert.
Then it seems like Autopilot’s Autosteer stayed locked on the left line even though it became the right line of the ramp. The system most likely got confused because the line was more clearly marked than the actual left line of the lane….” Read more Hmmmm… Must see video. Again… DoTs.. Just good paint, please!!! By the way, a similar situation happened to me a few years ago at the branch of the Southern State Parkway into the Belt Parkway near JFK. I was cruising along, top down, got confused by the lane markings and ran into the branch point. Luckily there was no barrier, but there was a curb. Blew the front left tire as I went onto the grass median. Still hoping to catch my plane to Sweden, I jacked up the car, changed the tire as cars were rushing by me on both sides. Went back onto the ramp to the Belt Parkway only to realize I also blew the back tire (cars don’t have 2 spares). I didn’t make it to Sweden, but I was befriended and helped by the nicest complete strangers. A most wonderful experience that I will never forget. Thank you Lenny Jones and Rev. White. Alain
Autonomous Vehicles: A Guide For Policymakers
B. Feigenbaum, Mar 29, “Autonomous vehicles, one of the most discussed technologies in the transportation industry, promise significant benefits, especially with regard to safety. With the advent of smartphones and telematics systems, distracted driving is an increasing threat to road safety….
Given all of the unknowns, transportation agencies and governments are unsure how to prepare for autonomous vehicles. In the next five years, policymakers should focus on the intermediate effects, including a world in which autonomous and non-autonomous vehicles share roadways.
To safely and quickly hasten the development and implementation of autonomous vehicles, policymakers should:..” Read more Hmmmm…Part 8, Recommendations starting on Page 50 is the most profound, as it should be.
The rest is a pretty straight forward presentation of the state-of-thought on AVs. I disagree on a couple aspect…
1. continuing to use the SAE levels as the basis for classifying the different technologies. They are not a good basis because they don’t clearly correlate to the policy recommendations. Advances in Safety are dependent on quality and functionality of Safe-driving technologies. Because these are basically Level 2 technologies, the perception is that these are mature technology. What we are finding is that they don’t work very well. Public Policy focused on Safety should be focused on improving these technologies which are far from mature. Taking hands off wheels, feet off pedals and drivers out of seats (the higher levels) have little, nothing or even negative influence on Safety. Safety-oriented Public Policy should be focused on the development and deployment of technologies that prevent crashes. They let us make the driving decisions as long as those decision aren’t placing the car in a position where a crash is imminent. If so, these systems automatically take over the driving function so as to avoid the crash. They apply the brakes, they slow down, they don’t tailgate, they return to the center of the lane, they yield, they lead, they change lanes, … Whatever is necessary to avoid a crash. Once the threat of a crash has passed, they smoothly return control of the vehicle to the driver. Anti-lock brakes and Electronic stability control are mature technologies that take over, then smoothly return control, in certain situations. The basic purpose of Safe-driving technologies is to deliver similar automated intelligent crash-avoiding driving during all of what would have been crash-imminent situations. Auto ownership is not an issue, nor is travel demand, nor is taking body parts off of controls. It is all about disregarding what those body parts might be trying to do and having a computer do the right thing.
“Self-driving” is a technology layer that is built on top of mature, high-quality Safe-driving technology. That technology is focused on delivering comfort and convenience to the driver without degrading Safety. This technology is focused on the routine of driving, rather than dealing with emergency situations. Adhering to many, but not necessarily all of the multitude of traffic regulations in different jurisdictions so that the driver doesn’t need to is its primary purpose. These systems still need the driver present to be able to drive the vehicle in situations that are challenging for the automated system. Safety is at best constant. Since these cars need a driver, ownership remains the same; however, because comfort and convenience are enhanced significantly, both trip demand rate and well as trip length are likely to increase and increase substantially.
“Driverless” is a Safe-driving with the highest quality of Self-driving features that adheres to all of the multitude of traffic regulations in its driving operating domain. It functions as well without a driver as it would have functioned with an attentive human driver. Instead of an individual driver, these vehicles will be managed by a central controller as fleets of interchangeable vehicles that can be dispatched to serve individuals as well as groups of riders originating at about the same place at the same time going in the same general direction. This shared-ride functionality will lead to improved utilization and efficiencies. Thus the ownership model is a fleet owner rather than an individual. Use is by many rather than by the individual owner. Demand will be affected by pricing and service which may well be best in moderately dense, so-called “transit-oriented” land uses. Person Miles traveled are likely to increase while Vehicle Miles traveled are likely to decrease proportional to the propensity of ride-sharing. For this technology, Public Policy focused on making Ride-sharing attractive is critically important. Unless one looks at these technologies from the Safe, Self and Driverless perspective rather than the SAE level perspective one doesn’t easily appreciate how each of these technologies are fundamentally different from a Public Policy perspective.
Only one company in California has applied to test truly driverless cars
A. Brinklow, Apr 3, “Monday was the first day for truly driverless cars to legally operate on California streets—and yet not a single autonomous vehicle hit the roads without the usual human safety drivers behind the wheel. …In an emailed statement, DMV spokesperson Jessica Gonzalez tells Curbed SF:
The DMV has received one permit for driverless testing. The department has 10 days starting April 2, to let the applicant know if the application is complete. If it is deemed complete the application will be thoroughly reviewed.
Read more Hmmmm…It has to be Waymo. No one else is close (unless it is a low-speed shuttle, there is always uncertainty ). Alain
Experience Counts, Particularly in Safety-Critical Areas
Editorial, A. Shashua, Mar 29, “Society expects autonomous vehicles to be held to a higher standard than human drivers…. it feels like the right moment to make a few observations around the meaning of safety with respect to sensing and decision-making.
First, the challenge of interpreting sensor information. …To demonstrate the power and sophistication of today’s ADAS technology, we ran our software on a video feed coming from a TV monitor running the police video of the incident….” Read more Hmmmm…This is very impressive, but let’s hope that if one is relying on images that one is using an imaging system that is better than a “dash cam” display on a TV monitor. Uber MUST release the actual image/video sequences. Alain
Hands Off With Ambarella’s Camera-Only Self-Driving Car Tech
D. Cardinal, Apr 2, ” It’s conventional wisdom in the self-driving industry that lidar is a must-have. From time to time some have argued otherwise, but almost every serious effort currently underway includes at least one lidar — until now. Chip maker Ambarella, best known for creating the image processing systems found in many cameras, has introduced its EVA (Embedded Vehicle Autonomy) development vehicle built on its CVFlow architecture and a phalanx of 20 cameras powered by 16 of its CV1 image and vision processing SoCs. I was lucky enough to be one of the first reporters to ride in one, as the firm is now doing limited testing on public roads in Santa Clara, California.
Ambarella isn’t trying to go head to head with the car companies and planning to launch its own self-driving vehicles. Instead, EVA, which is built on a modified Lincoln MKZ, is a development and design showcase for what Ambarella can bring to those companies in the way of an autonomous vehicle solution based almost entirely on vision, though it does have a single front-facing radar to help with tricky weather…
General Manager of Ambarella’s subsidiary VisLab Alberto Broggi emphasized to me the importance of integrating the vision modules with the ISP (Image Signal Processor) in the camera. That’s possible because Ambarella as a company makes much of its living developing ISP systems for a variety of applications….After all, humans are an existence proof of the possibility of night driving without lidar or IR illumination…” Read more Hmmmm… See video. Very nice. Alain
How Nvidia’s RTX Real-Time Ray Tracing Works
D. Cardinal, Mar 29, “Since it was invented decades ago, ray tracing has been the holy grail for rendering computer graphics realistically. By tracing individual light rays as they bounce off multiple surfaces, it can faithfully recreate reflections, sub-surface scattering (when light penetrates through a bit of a surface like human skin, but scatters back as it goes), translucency, and other nuances that help make a scene compelling. It’s commonly used in feature films and advertising, but the large amount of processing time needed has kept it away from real-time applications like gaming.
Nvidia is aiming to change all that with RTX, its high-performance ray tracing library that can churn out detailed scenes at game-worthy frame rates. We’ll take a deeper look at what makes ray tracing hard, and Nvidia’s RTX…” Read more Hmmmm… David, thank you for clearly explaining. Alain
Self-Driving Cars in a City Like No Other
O. Cameron, Jan 10, “Voyage is bringing self-driving cars to a retirement community (and city) like no other: The Villages, Florida. With 125,000 residents, 750 miles of road and 3 distinct downtowns, The Villages is a truly special place to live. Today, we’re excited to announce that Voyage has started testing its self-driving fleet within the community. Beginning in early 2018, we’ll start rolling out a door-to-door self-driving taxi service to residents….
With so many miles of road, high-resolution mapping of the community is a crucial piece of the puzzle to solve before we can serve residents. We are thrilled to be partnering with CARMERA, a world-class provider of real-time, street-level intelligence for autonomous vehicles. We simply wouldn’t be able to move as fast as we would like without a partner like CARMERA, for what we believe is the largest deployment (by area size) of self-driving cars in the world.
CARMERA is mapping all 750 miles of road within The Villages, and will provide continuous updates of localization and navigation-critical data that Voyage self-driving cars will consume on the fly…” Read more Hmmmm…Congratulations Ro. Alain
Calendar of Upcoming Events:
2nd Annual Princeton SmartDrivingCar Summit
May 16 & 17, 2018
Save the Date
Application NOW OPEN
Become a Sponsor and Display your wares