2014-05-23
May 22, 2014
Update on my “997” Package
Since I walk to work, I haven’t had the opportunity to test my Mercedes “997 package” every day; however, it has assisted me around New Jersey’s jug-handled roads as well as the NJ Turnpike, Garden State Parkway, I-195 and 295 as well as longer trips on the PA Turnpike, I-81 and I-95 and earlier this week to and from Cleveland on I-80.
The really good news is that the intelligent cruise control (Distronic) part of 997 works really well. To and from Cleveland on I-80 I used my feet only when I wanted to and not once when I needed to. (Of course, on non-exclusive roadways I needed to in order to obey traffic control devices). Also, my return trip was through some pretty nasty rain and fog and I was amazed at how well it “saw” the cars and trucks directly ahead of me. It may well have seen them better/sooner than I saw them. I liked the way it slowed down when needed; the return acceleration was a little more sluggish than I usually drive, but it is right, I’m wrong. The stop & go (Jam Assist) is really just fine and much more relaxing than when I do it. What I really liked was the way it responded to someone cutting in front of me. Except for the blowing of the horn, just the way I do it. With 997, speed control on limited access roadways, rain or shine, becomes: Set the speed at “9 over”, pick a lane and let her rip!
My only complaint about Distronic is how it seems to turn off when I touch the brake pedal. It doesn’t turn off if I step on the gas. It just says, OK I’ll step aside until you take your foot off the gas, then I’ll do my thing (and, I hope, if I, the 997 package, determine that a crash is imminent unless I take over, and takes over. I sure hope it does that! Have yet to experience it and have been too chicken to try it; although, I’ve thought about it. I’m too chicken!). However, when I touch the brake, Distronic tells me that it turned itself off. If I now take my foot off the brake it doesn’t turn back on. (I’m assuming that the emergency braking system is always on and always dominates Distronic. Again, I’ve been reluctant to test it.) What I found myself doing is: starting to brake before 997 as I approached a slower moving truck. Once I touched the brake, 997 now makes me responsible for not getting too close to that truck I’m following. That is, until I’m so close that the emergency brakes take over (I hope). Well that wrong; shouldn’t happen that way. Even if I touch the brake, then let off, 997 should intelligently follow that truck ahead instead of leaving it up to me.
That’s really unfortunate and a short-sighted holdover from un-intelligent cruise control. I’ll argue that Distronic should be turned off only if the user turns it off using the on-off switch. It should continue to compute as if the driver had never touched the brake or the gas. If what the driver is commanding through the pedals is different than what Distronic computes, then:
if the driver is accelerating, that wins over Distronic (but not over the emergency braking).
if the driver is braking, then the largest braking force should be implemented (driver’s, Distronic’s, Emergency’s)
Distronic should always be on.
The OK news is that the lane-centering sort-of works. To me, lane centering is very important. I really want to stay in the center of my chosen lane when the availability of the lanes to my left and right are comparable, but favor slightly one side or the other depending on some availability measure of the road surfaces beyond the edges of my lane. I have the feeling that 997 knows the lane boundaries sufficiently far ahead of me to properly steer this instant as well as the next few seconds ahead, but that the feedback control system that MB has implemented is “challenged” for reasons that are beyond me. I doubt that the source of the challenge are the sensors, the fusion of their data streams or the estimation of the lane boundaries. It has to be in the choice of feedback control model or in the way it has been coded. They must have better code and I certainly hope that they’ll send me a message soon that updated code is available for download so that the lane centering will now work. (They are going to do that aren’t they? I didn’t pay the big bucks for a system whose software was frozen in 2013. Please someone tell me that’s not true. Think of it… you bought your laptop in 2004, its hardware is 2004 hardware but at least you can run Windows 2013. OK, that wasn’t the greatest example. :-) ). Seriously, I believe with just a software upgrade many of the limitations of the lane centering system should be fixable and MB should properly treat its customers by making it available at a fair price!
I have complained previously about how lane centering shuts off if you should not immediately heed 997’s alert to put your hands back on the steering wheel. Please don’t punish me by simply turning off the lane centering even though you know the lane edges and could just as well continue to steer. This is no way to teach me a lesson. Please rethink this.
My biggest other complaint is about the UI. I understand that I am not a typical MB customer. However, since MB already gave me at least 6 different UI choices, please give me a 7th, one that looks more like the UI when I am backing up. 997 should display its view of the road ahead with all of the objects displayed that it has detected and that it is considering in its control of the brakes, throttle and steering. MB does this in their test cars; they have the UI software to do this. So why don’t they include it in their next software release to me? (rhetorical question) Please don’t think I’m too stupid. I’m a customer! I’d like to understand what the 997 package is thinking about when it decides to do what it has decided to do. All the data are sitting there in your memory. 997 should just use its nice display tools to share those data with me. It should be able to do it at essentially zero incremental cost. And, if I don’t want to see it, I don’t have to. I can select one of the other 6 displays options. So why won’t MB include this in its next software upgrade? (Please don’t say NHTSA won’t let MB do it or that MB doesn’t do software upgrades.) Alain
ps. The 997 package accomplished all of the above without accessing a single byte of Google or Bing or Nokia digital map data. I, through CoPilot Live, did refer to ALK digital map data.
Object detection software to enhance driver awareness, improve safety
By Colin Jeffrey, May 20, 2014 “… Fujitsu Semiconductor Limited is set to introduce software that assists in detecting and identifying cars, people, and other moving objects and alerts the driver of their position and direction of travel.
Using what Fujitsu calls its “Approaching Object Detection Library” in conjunction with a proprietary System-on-Chip (SoC) graphics device, the system uses the images captured by additional vehicle cameras to automatically detect nearby moving objects and compare them to stored representations to identify and categorize them…” Read more Nice to see Fujitsu in the game. Alain
DMV Adopts Autonomous Vehicle Testing Rules
May 19 “ The California Department of Motor Vehicles (DMV) today announced that the Office of Administrative Law has adopted regulations governing how manufacturers can test autonomous vehicles on California roads, effective September 16, 2014.
The regulations set rules for vehicle testing, insurance, registration and reporting by manufacturers. Today’s regulations only address manufacturer testing requirements; they do not address rules for operation of autonomous vehicles by the public. Rules governing public operation of autonomous vehicles are currently being developed by DMV and are expected to be adopted by January 1, 2015. Read more This is important. Alain
May 19, 2014 Self-driving cars pushed closer to reality in Europe
Reuters, May 19, 2014”…A little-noticed amendment to the United Nations Convention on Road Traffic agreed last month would let drivers take their hands off the wheel of self-driving cars. …The amendment agreed last month by the U.N. Working Party on Road Traffic Safety would allow a car to drive itself, as long as the system “can be overridden or switched off by the driver”. A driver must be present and able to take the wheel at any time…” Read more This is a substantial step forward! Alain
The Trick That Makes Google’s Self-Driving Cars Work
Alexis C. Madrigal May 15 2014, “…The key to Google’s success has been that these cars aren’t forced to process an entire scene from scratch. Instead, their teams travel and map each road that the car will travel. And these are not any old maps. They are not even the rich, road-logic-filled maps of consumer-grade Google Maps.
They’re probably best thought of as ultra-precise digitizations of the physical world, all the way down to tiny details like the position and height of every single curb. A normal digital map would show a road intersection; these maps would have a precision measured in inches.
But the “map” goes beyond what any of us know as a map. “Really, [our maps] are any geographic information that we can tell the car in advance to make its job easier,” explained Andrew Chatham, the Google self-driving car team’s mapping lead.
“We tell it how high the traffic signals are off the ground, the exact position of the curbs, so the car knows where not to drive,” he said. “We’d also include information that you can’t even see like implied speed limits.”…” Read more Big question… is it even necessary? Alain
Oshkosh works to develop driverless trucks for Marine Corps
Rick Barrett, May 12, 2014: “…The truck manufacturer is working with the U.S. Office of Naval Research to produce an unmanned vehicle that would be used by the Marines Corps in supply-line convoys, including trips into combat zones…With improved sensors, now they’re “seeing” in 3D, and they’re better able to tell what’s in the road.
That’s a big improvement from 10 years ago when, in a desert race in Nevada, a 16-ton Oshkosh robotic truck earned the nickname “gentle giant” after it balked at some tumbleweed, froze in its tracks and backed into a ditch….”
Read more That is nice progress. Alain
Half-baked stuff that probably doesn’t deserve your time:
Driverless cars could cripple law enforcement budgets
This is worse than half-baked. Automated cars will make traffic cops irrelevant and unemployed! Alain
GM partners with Google for ride-sharing pilot program
And why aren’t they using their driverless cars??? Alain
Self-driving Cars an $87 Billion Opportunity in 2030, Though None Reach Full Autonomy
Half-baked mostly because it is a private report that has only been distributed as press release headlines, which are somewhat questionable without the underlying details. The headlines:
China overtaking US & Europe. What else is new???
Software will be the competitive differentiator. Again what else is new? Although, one would really like to know what supports their estimates of a $25B software business in 2030. Assume in 2030 there are 100M households/automated vehicles. That’s $250/yr for software (in 2014 dollars). I doubt it! I thought it might include data, but their slice for data in their chart is so thin you can’t see it. So??? I disagree, and I imagine Google disagrees, that data opportunities are infinitesimal.
True autonomy remains elusive. Depends on the definition of “true autonomy”. What CityMobile2 is beginning to test some view as “true autonomy”. so again, without the details, this is half-baked.
Since these are the only 3 headlines in the press release, I then must conclude that this contains very little that is new over and above what, for example, Rand, KMPG and Lloyds have put out recently that are available in the public literature. Alain
Calendar of Upcoming Events:
2014 IEEE Intelligent Vehicles Symposium
June 8 - 11, 2014, Dearborn, Michigan, USA
Sponsored by the IEEE Intelligent Transportation Systems Society
http://www.auvsi.org/avs2014/register
Invitation ONLY
Save the date
September 6th, 2014 in Detroit
Recent Versions of:
#
May 22, 2014
Autonomous Vehicles: Handing Over Control: Opportunities and Risks for Insurance
Gillian Yeomans, Lloyd’s Exposure Management “…The insurance industry’s expertise in risk management will be a factor in the adoption of autonomous … technology. In an area where regulation and safety standards are yet to be developed, insurers can encourage prudent progress by making their own risk assessments and providing policies for responsible operators. There is an opportunity for insurers to engage in the transfer of new risks, making it possible for continued technological innovation. This technological innovation may give rise to new business opportunities, with corresponding opportunities for insurers….“Read more Nice to see Lloyd’s weigh in here. Alain
May 2, 2014
Who Is at Fault When a Driverless Car Gets in an Accident?
John Villasenor Apr 25 2014 “It may seem odd, but we already have the laws we need for dealing with this inevitable situation.” Read more especially the source report:
Products Liability and Driverless Cars: Issues and Guiding Principles for Legislation
April 18, 2014
Newly Available Drafts of Recent Student Theses
Wyrough, Alexander Penn Hill Jr.: A National Disaggregate Transportation Demand Model for the Analysis of autonomous Taxi Systems
Included is one synthetic realization for each of the 308.7 Million individuals contained in the 2010 Census and each of the 1.01 Billion trips {oLat, oLon, oTime, dLat, dLon} these individuals were synthesized to have made on a typical weekday in October. Persons are organized in individual Home County Files (All 186,49 persons that lived in Peoria County, Il., or the 1,585,873 persons that lived in Manhattan, for example). Trips are available in files for each person residing in a county (with many trips diffusing into other counties, states and even countries) as well as by oCounty (the 649,781 trips that originate in Peoria County or the 8,085,055 trips that originate in Manhattan, many of which are made by 1.5 million persons that don’t live in Manhattan.). For example, one can get all the trips made by the residents of Manhattan or by all trips originating in Manhattan. The user is left with the task of finding all trips that either originate, terminate or go through Manhattan. Note, trips to work where the work place was greater than 200 miles from the home location were routed via the nearest major airports. Alain
Sun, Zhuyi (Judy): Causal Price Discrimination: An Analysis of the Healthcare Costs Associated with Motor Vehicle and Transportation Collisions
Rhodes, Brandon Xavier: An Analysis of Economically Efficient Insurance Schemes for Automated Vehicle
April 11, 2014
Hands-On: 2015 Hyundai Genesis Flirts with Autonomous Driving
By Scott Schaen on April 7, “…Lane keep assist worked very well if you’re slow to make a slight turn. It doesn’t work for the bigger turns. Similarly, smart cruise control and the automatic emergency breaking system would slow you down a lot, but it won’t bring you to a complete stop. They’re SAFETY features; they’re not intended to auto-drive (yet)…” Read more
April 4, 2014
Webinar on the 3DV product
Thursday, March 27, 2014 Video of Webinar. Excellent presentation by Prof. Alberto Broggi Alain
March 28, 2014
Millennials & Mobility: Understanding the Millennial Mindset
“…millennials, with their relative propensity for urban lifestyle components (whether they live in cities or in suburbs), dexterity with technology, while starting careers during economically constrained times can leave a lasting impact on society….” Read more
March 21, 2014
Toyota Is Fined $1.2 Billion for Concealing Safety Defects
By BILL VLASIC and MATT APUZZO MARCH 19, 2014 “Eric H. Holder Jr., the United States attorney general, talked in impassioned tones on Wednesday about Toyota’s behavior in hiding safety defects from the public, calling it “shameful” and a “blatant disregard” for the law. A $1.2 billion criminal penalty, the largest ever for a carmaker in the United States, was imposed. Read more
Hmmm… As I wrote last week with respect to GM, the industry cannot respond in this manner during the evolution of SmartDrivingCars. “Defects” are inevitable, as will improvements to the systems. In fact, instead of a “defect recall”, the industry might create a “performance enhancement” recall process where software upgrades and some hardware upgrades may be made to the systems to improve their performance. The industry might/should be able to charge for these improvements. This would be like installing new and better “wiper blades”. Not that the old ones didn’t work, the new ones work better. And the industry could charge for it. A whole new revenue source. Your SmartDrivingCar could get a real “tune up”. :-) Alain
March 14, 2014
303 Deaths Seen in G.M. Cars With Failed Air Bags
By DANIELLE IVORY and HILARY STOUT MARCH 13, 2014 “As lawmakers press General Motors and regulators over their decade-long failure to correct a defective ignition switch, a new review of federal crash data shows that 303 people died after the air bags failed to deploy on two of the models that were recalled last month…Read more
Hmmm… This is tragic for many reasons; however, we in the SmartDrivingCars world need to learn very clear lessons… More importantly (and as is obvious to the most casual observer) we can’t wait for “…303 victims…” of the “we didn’t think of that” problem to occur before we fix it…. Alain
March 7, 2014
http://www.apple.com/ios/carplay/
Hmmm. Just what we need in our cars, an easier way to be distracted from driving. Maybe this should be called “Apple CarCrash”. Alain
New Jersey Rail Ridership Opportunities If aTaxis Are Available…
A.L. Kornhauser, et al March, 2014 “…. Examined are the rail ridership opportunities that NJ Transit (NJT) might enjoy if autonomous Taxis (aTaxis) were available to readily bring customers to or from its rail stations. Such collection and distribution services would not only shower NJT with new customers but they would enable NJT to redevelop its parking facilities to more profitable “Transit Village” uses. Read more
This list is maintained by Alain Kornhauser and hosted by the Princeton University LISTSERV.