Brad Ideas - Robocars - Comments

This "radar" doesn't so readily identify

brad — Fri, 21 Nov 2008 23:17:26 -0800

You get a resolution of about 1 inch, so this is not an ideal tool for identifying dogs, people and trees. Existing systems are pretty good at identifying other cars with it, both because of their shape and the fact that they are often moving. Ditto people and animals.

The one thing it does is give you very reliable assurance there is nothing in your path or other places you are interested in. If the LIDAR gives you a distant return you can be pretty sure there is nothing closer.

Once it tells you where objects are, other systems — including cameras and higher resolution, spot-based LIDAR, will be used to zoom in on the objects and identify them, as to what they are and what they are doing. If they are people, I expect cameras to zoom in on the face and determine where the person is looking — there are already systems that do this.

Right now the big problem camera based systems face is lighting. They can work in some lightings but not in others. LIDAR provides its own infrared light, so it works in all lighting.

I expect a robocar to feature LIDAR, wide field stereo cameras in all directions, actual radar (including down at the bumpers for parking) and narrow field zoom cameras on fast servos that can look at objects and people to identify them and their intentions. And anything else we think of.

Car Radar

Randy — Thu, 20 Nov 2008 22:26:58 -0800

I've been thinking about this for decades myself...but not as formally as you and these other researchers.

The radar would also have to tell the difference among a tree, a dog, and a person.

I can't wait for these radars to improve enough for a bumper to bumper stream of cars to drive 100 mph through an intersection. It should be quite scary at first :-)

Randy

Hmm... awesome about getting

M@ Childress — Thu, 13 Nov 2008 08:11:14 -0800

Hmm... awesome about getting to sit in a Tango, bummer that it wasn't really comfy -- Der Twike is a true two-seater that I've slightly modified into a 2.5 -- added a car seat. The Tango probably has a better crash protection tho' (protection from other vehicles). I've altered my driving habits from the full car -- stick to tertiary roads where traffic is a lot saner (same roads I'd bicycle on).

Yep, the video of the Tango carving up that course is very impressive -- shows that a floorboard full of batteries gives you a very low GG... the downside is that it's the weight of a small Camry (2-3,000 pounds if I recall). This becomes pretty wasteful in stop-n-go traffic or stop-sign-every-block style city driving, because as any bicyclist can tell you, most of your energy is spent getting moving from a dead stop... of course going towards inteligent cars that could sense traffic conditions, it's not as big a deal... unless you crash and you've got all that kinetic energy to disperse somewhere.

Back of a Tango

brad — Wed, 12 Nov 2008 12:57:14 -0800

Yes, I’ve sat in the back of a Tango. It’s not a great experience, but I should note about that.

Your Twike uses expensive nicads and has a 50 mile range. My vision of the mini robocar uses whatever battery is cheapest (including the cost of energy to move the weight of the batteries.) That’s probably lead-acid. As to how much is in a typical car, a subset of cars would be long range (50 miles like your Twike) but many would be only 10-20 miles. The question is the economics of downtime for the vehicle. If the vehicles and batteries are cheap, you give them less range and they spend more downtime charging (though at fast-charge stations.) If the vehicles are rarer and more expensive, you give them more range so they don’t have to recharge as much, or more probably you enable them for battery swap.

Lead-acid actually has an advantage. A tray of lead-acid on the very bottom of the vehicle provides stability.

The Tango is an in-line TWO seater EV...

M@ Childress — Wed, 12 Nov 2008 11:00:13 -0800

If you're talking about the Commuter Cars Tango in your charts, it is an inline TWO passenger electric vehicle (ie, one-behind-the-other, similar to motorcycle-style seating) . It is also illegal on the roads in most states (as it has 4 wheels it is considered a passenger car, and threfore requires crash testing to be anything other than a kit-car). That being said, I'd love to have one ;-)

I love your anlysis, of course that could be because I own and drive a 10-year-old Twike (Twin Bike), a three-wheeled Electric Vehicle/bicycle thingy ;-) www.illinois.edu/goto/twike

In the first year of ownership I put almost 4,000 miles on it, driving it around town, back and forth to work, to drop off Maddy-n-Mia (7 and 2 years old) at school and daycare, tasks that I could not accomplish (and get to work by 8:30-9am and school/daycare pickup by 5:30) using C-U's "Best little transit system"... going on year 2, and while the novelty has worn off, it's hard to have a bad day when people honk, wave and smile at you on the way home ;-)

The Twike is listed as getting 250-600 MPGe, and at the '08 Madison WI Hybridfest 26-mile test loop I calculated 168 MPGe on coal-generated electricity (which soars to a 500+ MPGe on renewable-sourced electricity) -- based on the MPGe formulas at Wikipedia.

Mass transit too

galacsi — Sun, 09 Nov 2008 06:38:22 -0800

Interesting ideas. And Electric bicycles and scooters are doing fine these days.

But you know mass transit can be modernize in the same way. With variable length trains made of electric modules able to recover some energy when braking.

A big advantage of mass transit is the separation of traffic. You should put all the containers on train instead of putting them on lorries.

Does not serve children, or the old, disabled and infirm

brad — Thu, 02 Oct 2008 01:28:47 -0700

You could not get a transit system like this today (even if it were a good idea, which I am not sure it is) because it doesn’t deal with the disabled, or others unable to walk to the e-bike depot or ride the e-bike.

It also does not work for shopping. To work in bad weather, you would need not bikes, but some sort of sealed vehicles.

What about this kind of system instead of robocars or PRT's

jim moore — Tue, 30 Sep 2008 20:38:53 -0700

For intra city transport I think that the key is to use a hub based transport system. Every type of transportation technology has distance range over witch it best suited. I would like to suggest that for intra city travel 3 distinct modes of movement should be used. Walking is ideal for distances of 1 to couple of 100 meters ( now of course walking for much longer distances is easily done just more costly in terms of time.) Next comes bikes / electric bikes / electric scooters / segways? which are good over the distances of ~100 meters to ~ 5 kilometers (~ 3 miles) (again you can go further but it becomes more costly in terms of time) Finally you get buses which would be used for distances of ~ 5 kilometers to ~ 100 kilometers (the width of the city).

Depending on density you have a transport hub in the center of a circle with a two to five kilometer radius. You have enough hubs so that everyone in the city is at most 5 kilometers from the nearest hub. Buses only travel between hubs and they only pick up and drop off people at hubs. To get from from the hub to your destination you are a member of a bike / scooter / segway share program that has big depots at every hub and little depots on just about every block.

So a commute would go something like this; you walk out of your house to the e-bike depot at the end of the block. You hop on a e-bike and ride the mile to the nearest hub (time from house to hub less than 5 minutes). You drop off the e-bike and hop on the bus to the hub nearest your place of work ( with fairly regular travel patterns it might be possible to optimize the the bus system so that you go directly to your destination hub without stopping at other hubs along the way). At the destination hub you pick up a different e-bike and ride to your work place which has an e-bike depot.

I think this type of system would be efficient, fast, flexible and inexpensive.

About rails

brad — Tue, 30 Sep 2008 11:29:22 -0700

I will also add that steel on steel is typically quite noisy. While pneumatic tires are slightly less efficient in rolling resistance, they make a smoother and quieter ride.

Can't simulate yet

brad — Tue, 30 Sep 2008 11:26:14 -0700

All we could simulate today would be not very smart algorithms. For parking lots we could probably simulate. I don’t think the algorithms used by Valets in valet parking lots show any particular genius. When I got to the valet lot, they ask me when I plan to leave. Even when I am off by quite a bit, it’s never much of a problem, and they pack those lights so tight you can barely walk between the cars. Part of the algorithm is to make both the cars leaving soonest and the cars with high uncertainty less blocked than cars with higher certainty.

Different from the Valet world, I expect the user to signal on their cell phone when they wish to leave, and that this will often take place several minutes before they get to the street. Most people have this sort of advance warning leaving offices, restaurants, parties and shops. Morning commuters know when they are finishing their breakfasts, and they usually have a pretty close idea of departure time in any event. In many cases events will be on the person’s digital calendar. Even if the calendar is wrong and the event runs late, as is often the case, it just means their car is more available than it needs to be for a while, no big whoop.

On a street, first we would see a lane of cars that fills the street. When it is time to add a 2nd, double parked lane, the parking controller computer on that street (just a local computer though it no doubt talks to the a more general parking controller) would look at the estimates and uncertainties and do the double parking as follows. It would find the car in the curb lane which is leaving soonest, and tell that car to move to a double parked position. The newly arriving car, which is most likely leaving later, would then take the vacated spot. As new cars arrive to double park, in most cases a car leaving sooner would vacate a curb spot and move to the double parked spot. If somebody needs their car in a hurry, then at most they just have to wait for the car that is blocking them to move out temporarily.

In fact, in the vast majority of cases, the double park lane will not be full. It will have at least one gap in it. In fact, the parking algorithms might decide to always leave one gap in all lanes, except the curb lane, even if doing quadruple or triple parking.

If there is a gap (long enough for the longest car stored inwards) then a whole chain of cars can move together, forward or back one space, to move the gap so it is positioned for the car that needs to leave. Then they can move again so that the car with the longest predicted stay can move into the vacated spot, it can be kept vacant and another car can enter.

This algorithm allows very dense parking and in fact if you leave these gaps you don’t even need to prioritize the cars by departure time because any car can get out in just a few seconds no matter how deep. It took me just a few seconds to think of this, and I am sure that people looking at the problem harder will find solutions even faster.

I am not endorsing the status quo. I am just saying we have enough, and robocars will give us more, and as they do, we can convert old parking lots to something nicer, or add more pedestrian areas to streets and so on.

PRT guideway can only be a robocar conduit if it’s more general, like ULTra. Many planned PRT guideways, such as suspended monorail guideways, steel truss rails etc. would require very special robocars for use. While people would probably buy such cars if it got them a faster trip, there would be extra cost.

Have you considered writing simulations to demonstrate?

Mike C — Tue, 30 Sep 2008 08:16:12 -0700

I think it comes down to a basic disconnect between our assumptions: you believe that automation will have a large effect on capacity (both street and parking capacity), whereas I believe that the effects will be limited.

Do you have simulation experience? It would be interesting to see a robocar world in action using simple 2d grid animation. My own personal skepticism is based on a belief that the combination of limited space and the requirement to mix with pedestrians will ultimately limit what you can do in the street, especially considering that some of the most difficult issues are most prominent at the busiest times. I would be less skeptical if you could demonstrate these gains in simulation (which should be achievable, since I'm concerned only with dynamic movement, not detection and identification of obstacles).

But my gut feeling is that privately owned robocars will not solve the capacity problem, in traffic or parking. A few other points in direct response to your last post:

vehicle arrangement - people are unpredictable. I think it will be difficult to predictively arrange double-parked vehicles in a way that minimizes the need to move outer vehicles to release inner ones. Again consider 3pm, the beginning of the rush: you have high pedestrian movement, a large amount of vehicles parked, and a traffic surge, all at the same time that the need to get "buried" vehicles is at a peak. I think this worst case scenario will ultimately limit capacity. I also disagree that the vehicle movement will not be disruptive - if they are really packed in, vehicles will frequently enter traffic to allow an interior vehicle out - again, a simulation would help to resolve this question

"clever grad students designing algorithms" - algorithms do well when input is predictable, but humans are inherently unpredictable. Perhaps you could make it part of the human-robocar interaction, where the human actually tells the robocar how long he expects to be? That could work, I guess, but people change their minds. A system couldn't rely on these predictions (what happens when the football game goes to OT and 40,000 people are 1.5 hours later than expected?)

"we already have more than enough parking for the current car load, by definition" - yes, but at what expense? How many old city buildings have been torn down for parking? I don't view the status quo as acceptable here. Our cities have been decimated and pockmarked by the automobile - we need to work to reverse that trend, not introduce new technologies that will maintain or enhance the trend. And I remain skeptical that robocars will provide anything more than incremental capacity gains using only the currently allocated space to cars, meaning that a robocar world would merely sustain the current trends, albeit at a lower pace (simulation could prove me wrong here)

"taxis don't vanish" - actually, for all intents and purposes, taxis do vanish. If a taxi makes 20 trips per day (2 per hour over 10 hours), and assuming people make round trips, that's 10 private autos removed from the street for every single taxi. That makes the taxi 1/10th the parking space cost of a private vehicle. Shared robocars and PRT would get the same benefit. This is the basis for my insistence on shared vehicles.

subway tunnels as "express robocar conduits" - this is precisely what I would suggest for PRT guideway - it can become an express robocar conduit in the future, therefore not wasted

The bottom line, I think, is that we have differing assumptions. I feel that robocars will not provide all the gains you claim they will, and even if they do, I am concerned that they will continue the current trend of clogged streets and density-destroying parking lots. I'll keep an open mind on robocars, but I won't abandon promising solutions today based on something that won't arrive for 30 years, if at all.

Not how valets work

brad — Mon, 29 Sep 2008 12:38:02 -0700

Yes, on-street parking is more efficient than parking lots because the street is the access, but it’s not as efficient as you think. In particular, the presence of driveways, hydrants and the rest constrain things, leaving some gaps no modern car fits in, and other gaps not quite long enough for two. On average, every inter-driveway gap will waste half a short car of space, but those gaps are wasting much more.

Valet lots always get a rough idea of when people will want their cars. They arrange the cars so that those cars leaving first are on the outside, no matter when they arrived. Triple-parked robocars will do this, but actually do it a lot better as it is no “work” to do rearrangements when needed, just a relatively tiny amount of energy.

A large fraction of parked cars have a roughly scheduled time they will be used again, especially when parked for work, or meetings, or movies or even meals. “I have no idea how long” is a minority of the cars. I expect clever grad students to work out wonderful optimal algorithms for arranging them.

However, the other reason there is not a problem is that we already have more than enough parking for the current car load, by definition. (Yes, the price of parking is enough to drive some to transit or carpools, and those people will be the first to switch to robotaxis or keep using transit.) So when you start with enough parking for today’s cars, and you add the huge increase that robocars can bring, I am not worried. If areas get denser (instead of less dense as I am tentatively predicting) the market for parking will sort itself out, especially if cities stop providing free parking.

New York is the most challenging place in the USA, but I think it will work. All those taxis don’t vanish when not doing fares. They are either cruising for fares or standing somewhere during lower demand times. A car cruising for fares takes more space on the roads than one sitting standing in a roadside space. Inherently you can handle more cars on the road stopped than you can moving. Again, if there is no demand for the vehicles, they just take over a new lane of a road, a double parked lane. It does not hurt that they took away that lane of capacity because obviously we are at a low demand time or these taxis would be at work moving people.

The subways in New York won’t go away, of course, at least not the tunnels. At the very least they will be turned into express robocar conduits, or may well still run at rush hour.

Robocar parking flexibility is not a panacea

Mike C — Mon, 29 Sep 2008 11:19:58 -0700

I do agree that robocars will more effectively use parking space, but it will not solve the problem completely because you still need space to park all those tens of thousands of privately owned robocars. In a uniformly dense city you'd have to bring the robocars all the way back out to the edges of the city or into the suburbs to find space - or keep a significant portion of those lots in the city core.

Parking today is pretty efficient in terms of space usage per vehicle, and I don't think robocars will significantly improve on that. Parallel parking probably only wastes 10-20% of vehicle space; for lots and ramps, it's probably more like 40%. Even if a significant portion of robocars are half-sized vehicles and they are able to use 100% of the space, I don't think you could do much better than perhaps doubling parking capacity in the currently allocated space.

And I don't see double and triple parking as workable for privately owned vehicles, because every time someone leaves you'd have to reshuffle the outer vehicles. That means potentially 2-4 robocars moving out of their spaces to allow an interior private robocar out, and those 2-4 robocars will now have to find another spot, or circle around and re-park.

Now consider the beginnings of the afternoon rush: those on 6am-3pm shifts will require their cars right at 3pm, but their robocars are buried 3 cars deep behind robocars from later shifts (7-4, 8-5), so you'll have a large amount of vehicle shuffling at the very time when you need the most capacity - during the rush.

For these reasons, I think private ownership of robocars should be, at best, a premium option with much higher usage and parking fees. If private vehicles are to be allowed at all, they must be priced to reflect the additional burden they place on the city relative to public vehicles which can each make 6-10 trips during the rush (thereby reducing the volume of parked cars by up to 90%). I have this same view of privately owned PRT/dual-mode vehicles: it should be discouraged or forbidden.

Take New York City as an example: you yourself said (on the TI group) that many rely on taxis for transportation in New York. Taxis are today's equivalent of tomorrow's publicly owned robocar from a functional standpoint (not an ownership standpoint, but we'll get to that in a minute). The reason taxis work in New York is that (a) the streets are largely kept relatively clear by underground transit, and (b) they don't require parking. Imagine if all those taxi riders in New York required private vehicle parking - it couldn't work.

And when I say "publicly owned", realize that I mean "shared, not individually owned". There could be privately owned "robocar companies" like cab companies that lease out vehicles and maintain them. This would remove the individual passenger-to-vehicle link that prevents vehicle sharing and complicates parking. With robocars, true door-to-door service with shared vehicles would be possible, as with taxis today.

So perhaps a robocar future with a high percentage of shared vehicles would alleviate the parking crunch, but not if most vehicles are privately owned.

But, again I reiterate, this is all 30-years in the future at least, so PRT investment today is worthwhile.

Parking

brad — Sun, 28 Sep 2008 22:05:32 -0700

Let me first address parking. Today, the city subsidizes parking greatly in several ways.

First, they provide free or cheap on-street parking for anybody on most streets.
They often provide free or cheap city owned lots.
They provide free parking to residents of congested areas by offering special permit stickers that allow long street parking.
The provide building codes that demand that houses have garages, or that new stores or residential towers have lots of parking. (Sometimes however, they do the reverse, and recently some cities have been insisting the residential towers have only 1 spot for 2 units.) Where parking is required, those without cars are still forced to pay for it.

Of course, we drivers love free parking.

In the robocar world, I think parking problems go away. I should write a special piece on this:

Just about anywhere is usable as a parking spot because by today’s definitions, robocars never park, they just stand. They are never driverless. They can happly stand in front of fire hydrants or blocking driveways. On command, they move out of the way.
Likewise, robocars can double park and triple park, and again, move out of the way on demand.
Robocars will always park at valet density, or even denser, in parking lots.
Robocars don’t need to park close to their riders destinations, and in fact can park quite far away if the next trip has sufficient advance notice or is in fact even roughly scheduled the way most commutes are.
When lots of robocars are stopped, by definition there is not much traffic. In these low traffic times, robocars can double park on the streets, narrowing them — which is fine due to the lower traffic. At rush hour, all robocars will be in use, and on-street parking will be forbidden to those not in use.
Smaller robocars, notably 1 person robocars, should be able to park tightly — three in a typical non-compact parking spot. Valet parked, perhaps 5 times as many vehicles as an old-style lot.
In the future, robocar-only parking garages could have 5’ ceilings, more than doubling capacity.
Alternately, cars like MIT’s proposed city-car, which can stand up vertically when parked via stacking or some jacks, could further increase density.

All that adds up to a lot of parking. So much that I would advocate giving up free parking, and instead selling parking in live auctions. I think it would be cheap, and if it got too expensive, robocars would just move outward to where it was cheaper. City owned parking could be sold with 2nd price auctions. 2nd price auctions have the interesting result that if there is only one bidder, the 2nd bidder’s price is thus zero, and the uncontested resource is provided for free, or some base monthly fee. As soon as capacity is reached, however, people pay, and the price goes up quickly to a fair market price for the contested results.

We’ll buy software for our cars to let them shop the market for parking. And if there is not enough parking, everybody will be getting in the game. People will rent out their driveways — why not, it’s not exactly doing anything when you’re not there — to vehicles that will swish out of the way before you even get home.

Not only that, lots will rent out their spare spots in a “spot market.” Today, each parking lot tries to be sized to meet the peak demand of the buildings it is servicing. Robocars can spread out the demand over a very wide area, so now you have a pool of parking lots all serving the entire area. Vastly more efficient than today’s system.

Finally, I think a lot of people will switch to robotaxis. Which don’t need to park at all in the traditional sense, though they do need places to wait during low demand. All this removes the parking demand from those who give up owning a car or robocar.

I think with all this, there’s so much parking available that we’ll see parking lots being demolished and turned into other uses, until the market price for parking goes up enough to justify using land that way again.

OK, I see your point...

Mike C — Sun, 28 Sep 2008 18:33:10 -0700

I now see what you're getting at - that optimum density in a PRT world may well be less than optimum density in today's transit-centric city. I think that's a good possibility, since there will no longer be a need to pack people in to make high density transit feasible.

My point was that PRT would not artificially reduce density with destructive infrastructure, as cars do. It may well be as you say, that PRT cities may be overall less dense than the typical transit-dense city (New York) is today. But it will be a density limited naturally by local preference, not imposed by the inherent density-reducing infrastructure required by the automobile.

And even though I generally support the idea of a robocar future, I remain concerned that an exclusive robocar-on-the-streets approach will continue the current pattern of neighborhoods being clogged by non-local vehicle movement and endless parking requirements. Public vehicles on elevated guideways solve the density issue regardless of whether it's human driven or robocars on the streets below them. Robocars and PRT are complementary, even synergistic.