2014-01-18
SmartDrivingCars_01714 TRB Review
January 17, 2014
Autonomous Vehicle Technology: A Guide for Policymakers
by James M. Anderson, Nidhi Kalra, Karlyn D. Stanley, Paul Sorensen, Constantine Samaras, Oluwatobi Oluwatola
This report is excellent and I, of course, concur on v2v and “DSRC”. While some are now suggesting that “connected automation” is the way to go I continue to press that it should be “automation connected” even though it doesn’t sound very good. Automation must come first. In fact, that makes it a very good sound bite because it emphasizes that connectivity only becomes practical if it is augmenting and advising an automated system, rather than human. Else, it will be both ineffective and often too disturbing that users will turn it off. In a true emergency, my automated system has a better chance than I to know what to do. (How many rollovers have been caused by drivers doing the wrong thing while trying to correct a skid? NHTSA imposed Electronic Stability Control because drivers knew what to do? I don’t think so!)
The report’s discussion of the NHTSA “Levels of Automation” prompts me to comment that it is inappropriate to tag any vehicle as simply being in one level (except those that are only Level 0). Even a Level 4 vehicle may need to operate as Level 0 at some times in some places under some circumstances. This is true of any automated system. Firemen have special tools to manually operate elevators during fires. Consequently, what becomes important is to define very well the operating regimes for each car. On what roads, during which times under what circumstances (weather, state-of-good-repair …) is Level x functionality sufficiently reliable for this vehicle. It will be important for the vehicle to know its limitations, be able to determine how best to operate while constrained by those limitations and properly communicate with its user or owner if, when and where it may need help. This is especially true for Level s 3 and 4. (Levels 1 and 2 are such that the user must assume that the systems are never working. Not a very compelling consumer opportunity!) Hopefully, car manufacturers will realize that there are many roads, during many times, under many weather conditions where their lane keeping and collision avoidance systems function extremely well and they will creatively develop a way to readily inform the driver that such an operating region exists on the route ahead. The car will inform the driver when it becomes OK to consume Level 3 functionality if he or she chooses: however, the driver will need to successfully regain control of the car prior to the real end of the Level 3 operating regime; else the vehicle will pull over and stop prior to the end of the Level 3 regime. This requires that places to “pull over and stop” be available near the end of essentially all Level 3 capable road segments. For now, that may constrain the extent of Level 3 capable roads. I plan to develop an app for that.
I can see it now… Verizon type ads will become common for auto companies: Can I text now? GM’s Automated Lane Centering now works on 76% of the nation’s lane miles. Ford’s jam assist relieves the stress of stop & go traffic on 82% of the nation’s most congested roadways. (I may have just finally lost it, sorry.)
My only real disappointment with the Rand report was that it essentially had no mention of transit and goods movement/ commercial vehicles. Otherwise, it is an excellent report. Alain
Telematics Industry Insights by Michael L. Sena
Vol. 1, Issue 2 Michael provides his insight to CES 2014 as other news about connected vehicles. Alain
10 Autonomous Driving Companies To Watch
By Chuck Tannert “…We’ve put together a list of 10 companies, all makers of advanced driver assist technologies or systems, to watch now…” Read more Hmmm… I agree with some on the list, but others seemed to be focused on TravelTainment. Nothing wrong with that, if they also focus on the solution, SmartDrivingCars, to the problem they are creating with Traveltaiment. Alain
Highlights of Recent Events:
This was the last year that “TRB” will be at the Shoreham + Hilton + Marriott Hotels. Thank goodness! Next year we’ll be spread throughout the Washington Convention Center on Jan 11-15, 2015. Be sure to “save-the-date”: 6:00 -> 9:00, Tuesday Jan 13, 2015 for the 43rd Annual Princeton TRB Banquet. Owen Curtis is going to find a restaurant with a private room so that we can be sure to go around the tables and give everyone the opportunity to share their highlights for the past two years. Unfortunately, this year, the seating arrangement made it impossible for anyone to introduce themselves. Regardless, a good time was had by all.
SmartDrivingCar activities were one of TRB’s “Hot Topics”; however, TRB completely underestimated how hot and failed to assign sessions to sufficiently large rooms. All sessions and committee meetings dealing with SmartDrivingCars were SRO (Standing Room Only).
Starting Sunday with the SRO Workshop on Travel Demand Modeling Implications of Driverless Cars through the Thursday morning session planning for next summer’s 3rd midyear Automated Vehicles Symposium to be held at the Hyatt Regency, San Francisco Airport, July 14 -> 18. Save the Date!
In the Sunday Workshop, I was tasked to “Set the Stage” with my presentation “Urban Planning & Community Design Considerations in an Era of Driverless Cars”. Excellent presentations were made by Prof. Hani Mahmassani of Northwestern and Steven Polzin of the U of South Florida. I hope to post links in the next issue of SDC.
On Monday I attended the Shared-Use and Ride-sharing activities, both of which are including a keen interest in how automated vehicles can reduce the empty repositioning cost of Share-Use as well as the Shared-Ride opportunities of autonomousTaxis.
Tuesday afternoon, Dr. Jerome Lutin and I gave our presentation on the “Application of Autonomous Driving Technologies to Transit”.
Wednesday morning, Dr. Stan Young and I presented a summary the Transit and Ride-Sharing Research Needs Statements that were derived from last summer’s 2nd Automated Vehicle Conference at Stanford.
Later on Wednesday I presented a summary of the work that my students and I have been doing over the past three (3) years trying assess the ride-sharing potential of autonomousTaxis (aTaxis).
Some envision personal driverless vehicles providing auto-like mobility from anywhere to anywhere at any time. Moreover, we know that unless one’s personal trip desires are personally correlated or otherwise coerced to be similar to another person’s trip desires, individuals will travel alone in their own private aTaxi. Average Vehicle Occupancy (AVO) will be 1.0 as it is with today’s automobile.
However, modest operational limitations might serve to aggregate trips spatially and temporally without seriously compromising the attractiveness of aTaxi service, even if it means sharing the ride with complete strangers.
Such spatial and temporal aggregation should induce substantial casual ride sharing as normally occurs in elevators serving tall office buildings.
Throughout the morning rush, workers enter tall office buildings from many doors destined to work areas scattered throughout the building. Conceptually, elevator service could exist from every door to every workspace so as to provide each worker with a non-stop personal elevator ride to his/her work space. If each elevator waited at the workspace until the end of the workday, then the system’s AVO would be 1.0; again, just as with today’s automobile. We all agree that such a system is totally impractical.
However, by creating a “centralized” bank of elevators, with each elevator serving a group of floors, users are coerced to spatially aggregate themselves by walking to the elevator bank and entering the appropriate waiting elevator. Temporal aggregation is achieved by requiring that the door stays open for short while, thus enabling others going to one of those floors to casually share the ride. The door then autonomously closes and, magically, the elevator autonomously travels to the appropriate floor(s). Workers exit the elevator and disperses themselves by walking to their offices.
This operational model is clearly very “robust”. AVOs will tend to be large in the dominant direction at the busiest times enabling a few elevators to serve many customers because everyone is comfortable sharing the lift. During these hours many customers may need to endure a few intermediate stops; whereas, at other times most customers will experience non-stop service.
Similarly, an aTaxi system operating from geographically dispersed (GD) aTaxiStands serving dynamically assigned “neighboring” common destinations (CD) would spatially aggregate trips. Delaying the departure (DD) of the first customer to wait for the arrival of casual ride sharer(s) accomplishes the temporal aggregation.
The basic question my students and I have been struggling to answer over the past three years is:
Given: a demand for travel; say, the mobility needs of all people 9+ million people who live and/or work in New Jersey on a typical weekday. (Click for trip files, person files. Use freely but cite properly.)
Determine: How far will users need to walk (GD), how many intermediate stops will need to be endured (CD), and how long will the door need to be kept open (DD) in order to create sufficient opportunities for casual ride sharing so that the cost per ride and societal consequences of the ride sharing are very attractive?
Assessment: Is that level of coercion sufficiently docile relative to the commensurate societal benefits of such an AVO, so as to make such a system socially desirable? (Draft versions of county results)
My presentation summarized our efforts to construct a sufficiently precise representation of each of the 32+ million trips that occur throughout New jersey on a typical weekday as well as the computation of AVO values for various state-wide aTaxi systems characterized by various values of {SD, CD, DD}.
Findings: By maintaining the existing NJ Transit commuter rail network, locating aTaxiStands no farther than a 5 minute walk to every school, work place and essentially every home and activity center (CD = 5 minute walk), limiting the maximum Departure Delay to 5 minutes (DD = 5 minutes) and dynamically serving common destinations that impose less than a 20% travel time penalty on any traveler requiring and no more than 2 intermediate stops (CD=3), a daily New Jersey-wide casual AVO above 2.0 is achieved for those 32 million trips (3M are served by walking or biking, 1.2M use NJ Rail at least part of the way).
Higher AVOs are achieved in congested areas in congested directions such that all roadway congestion is eliminated (Some aTaxiStands are very busy so that they will need to be creatively designed to efficiently marshal the customers out of and in aTaxis and aTaxis out of and in aTaxiStands.)
The cost of a ride, energy consumption and environmental impacts would be reduced by more than 50% relative to a personal vehicle system. Plus everyone would have mobility.
I see substantial societal benefits and modest personal financial benefits (inexpensive ride) imposed by minor inconveniences (short walks and, at times, “sharing a lift”) with a surrender of the burdens and pleasures of personal automobile ownership. Sounds like a deal! Too good to be true?? What do you think? Read more Alain
TRB’14 concluded on Thursday with Session 871 Road Vehicle Automation The focus of this session was to plan for next summer’s 3rd Vehicle Automation Symposium to be held in the Bay Area. To get everyone thinking, two debates were conducted. One 2020 vs. Much Later with Steve Shladover, University of California PATH Program vs. James Misener, Consultant and the other Connectivity vs. Autonomy John Capp, GM vs. Alain Kornhauser, Princeton Here are my opening remarks. It was a lot of fun. Alain
Driverless car technology takes center stage at Consumer Electronics Show
While this is one person’s interpretation, I had a slightly different impression of CES. Mercedes was the only car company that seriously focused on SmartDrivingTechnology. They had both a 2014 E- and S-class sedans equipped with the 997 lane centering package as well as the other Intelligent Drive features. On a large video wall they featured these SmartDrivingTechnologies much as they did at the Frankfurt Auto Show. The people working the floor spoke intelligently about these features.
I did not have a similar experience when I visited the other manufacturers. At the Chevy/GM booth most of the action was in their central display area that I was not permitted to enter because I did not have an appointment.
Company Demos Driverless Shuttle
“…French company Induct on Monday showed off the first driverless vehicle to be commercially available in the U.S.
The Navia shuttle isn’t ready for U.S. street traffic yet, but this standing-room-only shuttle can transport up to 10 people from point to point on university campuses or in airport parking lots at speeds topping out at 12.5 mph…” Read more Retirement communities should pay attention to this. Alain
Audi highlights self-driving cars and more while they opened their exhibit with a driverless Concept Car, similar, but less dramatically, as Mercedes had done at Frankfurt, I found them to be more focused and infatuated with the “coolness” of their laser headlights rather than any SmartDrivingTechnology. Alain
At Mazda they were most proud of how they had located their front seat display high on the dash where the driver could more easily glance at what was being displayed. Other than traditional collision warning they were not touting any active safety devices. They were still focused on how to make available as much travelTainment as possible while continuing to emphasize that the driver must keep his hands on the wheel and feet on the pedals.
BMW was still “the ultimate driving machine” with little signs of evolving to “the ultimate riding machine”. They did a demo with a prototype car on a track:
BMW hits the performance limits with its driverless car
by Wayne Cunningham “…I sat in the passenger seat while a BMW staffer sat in the driver’s seat. He kept his hands off the wheel and feet off the pedals as the car roared toward the turns, hit the brakes, and swung the wheel over….It may seem like cheating that, for this demonstration, BMW programmed the car’s route in through GPS, which gave it the path around the track.” Read more It was! Alain
As far as the aftermarket guys, they were all focused on travelTainment, big time! It is amazing the lengths that they will go to to develop gizmos that are aimed at engaging the driver’s visual and auditory inputs without regard to the extent that these features my degrade one’s ability to drive. They must all be confident that drivers can readily multi-task. That must be what is selling; however, however, if they bundled those gizmos with SmartDrivingTechnologies that drove safely on some roads and some times, then the driver could really consume and enjoy that travelTainment on those roads at those times. Alain
Nvidia Aims Its Supercomputer Chips At Self-Driving Cars
By Klint Finley 01.06.14 “…the chip maker unveiled its new Tegra K1 mobile processor at the Consumer Electronics Show in Las Vegas. The chip is based on the company’s Kepler architecture, which is traditionally used by the chips that power some of the world’s most efficient supercomputers, but the K1 consumes even less power than previous Kepler models, making it suitable for phones, tablets, and, yes, autonomous cars….The K1 will come in two varieties: a 32-bit chip and a 64-bit chip — Nvidia’s fist mobile 64-bit processor. While Nvidia’s existing Kepler supercomputer chips include 2,880 processors cores, the K1 will have only 192. But that’s enough, Shapiro says, to provide sophisticated processing on cars as well as smartphones. The K1 should be available for tablets, mobile phones, and the like during the first quarter of this year. But Shapiro says the automotive version will take some more time due to the industry’s rigorous certification requirements…” Read more Hmmm…Maybe… Seems like they are going after the travelTainment market, but it would be nice to have 64 bit, 192 core processors available. As Nvidia has found, parallel computations are the norm in the image processing world. They are needed in the SmartDrivingCar world. Welcome! Alain
Very interesting student PowerPoints and Drafts of Chapters. A little rough in places but still well worth reading. Alain
http://orfe.princeton.edu/~alaink/NJ_aTaxiOrf467F13/Orf%20467F13_FinalReport&PresentationLinks.pdf
Calendar of Upcoming Events:
Recent Versions of:
January 6, 2014
Self-Driving Cars Moving into the Industry’s Driver’s Seat
Jan. 2, 2014 “Accident rates will plunge to near zero for SDCs, although other cars will crash into SDCs, but as the market share of SDCs on the highway grows, overall accident rates will decline steadily”. Self-driving cars (SDC) that include driver control are expected to hit highways around the globe before 2025 and self-driving “only” cars are anticipated around 2030, according to an emerging technologies study on Autonomous Cars from IHS Automotive, driven by Polk.
In the study, “Emerging Technologies: Autonomous Cars—Not If, But When,” IHS Automotive forecasts total worldwide sales of self-driving cars will grow from nearly 230 thousand in 2025 to 11.8 million in 2035 – 7 million SDCs with both driver control and autonomous control and 4.8 million that have only autonomous control. In all, there should be nearly 54 million self-driving cars in use globally by 2035. Read more
December 27, 2013
December 20, 2013
“The New Killer Apps
How Large Companies Can Out-Innovate Start-Ups” by Chunka Mui and Paul B. Carroll Now Available Highly Recommended. See also Chunka’s Dec. 19 Forbes article Will The Google Car Force A Choice Between Lives And Jobs?
This list is maintained by Alain Kornhauser and hosted by the Princeton University LISTSERV.