Submitted by brad on Mon, 2010-12-06 09:41.
The folks at the SARTRE road train project have issued an update one year into their 3 year project. This is an EU-initiated project to build convoy technology, where a professional lead driver in a truck or bus is followed by a convoy of closely packed cars which automatically follow based on radio communications (and other signals) with the lead. They have released a new video on their progress from Volvo.
I have written before about the issues involved in this project and many of them remain. It’s the easiest way to get a robocar on the highway, but comes with a particularly high risk if it fails — and failure in the earliest stages of robocar projects is very likely.
In the video, some interesting elements include:
- The building of a simulator to test driver attitudes and reactions. Generally quite positive, in that people are happy to trust the driving to the system and the lead driver. This will change a bit in a real car, since a simulator can only do so much.
- The imagine people eating, drinking, listening to music and reading while in the convoys, but they don’t talk about the elephant in the car: sleeping. People doing anything else can quickly take the controls in a problem, but sleepers may not. And there’s also that act that we metaphorically call “sleeping together.”
- Their simulations depict cars leaving the convoy from the middle. However, in this situation it seems you can’t give them too much brake-accelerator control for the difficult task of changing lanes when you are just a few feet from the cars in front and back of you. You must maintain the speed of the train until you have fully left its lane, but that means you can’t do the usual task of changing speed as you enter your new lane. Exit from the trains will need some work. (There are suggestions in the comments that make sense.)
- They expect to have to make legal changes to allow this. However, since it’s an EU initiated project, they have a leg-up on that. This might pave the way for more robocar-friendly laws in Europe.
- While they plan to do a live test by 2012, they are much more cautious on predicting when the trains might be common on the roads.
- They do speculate if a simple robocar function for “stop and go” traffic, which is able to follow the car in front of you at lower speeds, might come first. Indeed, this is pretty easy, and not much more than a smarter version of existing auto-follow cruise control with steering and lane-following added.
- Their main pitch is environmental, as drafting should save decent fuel. However, I think most people will be interested in the time saving, and I’ll be interested in how the public accepts it.
Submitted by brad on Mon, 2010-11-29 15:28.
Two bits of robocar news from last week. I had been following the progress of the Stanford/VW team that was building a robotic Audi TT to race to the top of Pikes Peak. They accomplished their run in September, but only now made the public announcement of it. You can find photos and videos with the press release or watch a video on youtube.
This project began with the team teaching the vehicle to “drift” — make controlled turns while wheels are skidding, something needed on the windy curves and dirt/gravel/pavement mix on the way up to Pikes Peak. Initial impressions were that they had the goal of being a competitor in the famous Pikes Peak Hill Climb — a time trial race to the top by human drivers, the fastest of whom have climbed in in 10 minutes, 3 seconds in major muscle cars. The best standard cars have done it in about 11.5 minutes, and Audi says a stock TT would take a bit under 17 minutes.
The autonomous Audi’s time of 27 minutes, with a top speed of 45mph, is thus a bit disappointing for those who were hoping for some real man vs. machine competition. The team leader, Burkhard Huhnke, downplayed this, saying that the goal was to come to a better understanding of computer controlled cornering and skidding, in order to make better driver assist systems for production vehicles. Indeed, that is a good goal and it is expected that robocar technologies will first appear as driver assist and safety features in production cars.
The actual run was also marred by tragedy when the helicopter filming it crashed.
Earlier, I spoke with James Gosling — more famous as the creator of the Java language — about his role in the project. Gosling knows languages and compilers very well, and he helped the team develop a compiler so the interpreted scripts they were writing in languages like Matlab. Gosling’s compiler was able to run the resulting code around 100x faster than the interpreter, allowing them to do a lot more with less hardware.
There is strong interest in man vs. machine robocar contests. Such contests, aside from setting a great bar for the robots, will demonstrate their abilities to the public and generate strong public interest. This turned out not to be such a contest, but someday a robot will race to the top of Pikes Peak in better than 10 minutes. It will have a bigger engine, and many more sensors than the Audi in this run, which mostly relied on augmented GPS (extra transmitters were put by the roadside for full accuracy.)
A future car will have a complete map in its head of where all road surfaces are, and their characteristics. It will know the physics of the car and the road better than any human driver. The main thing humans will be able to do is use their eyes to judge changing road conditions, but they don’t change very much, and computer vision or sensor systems to make such judgments don’t seem like an impossible project.
Masdar PRT in operation
In other news, the greatly-shrunk Masdar PRT system, built by 2getthere Inc. of the Netherlands, has entered production operation in Masdar, an experimental city project just outside Abu Dhabi. The project only has 2 stops for passengers (and 3 more for cargo) at this point. It runs at ground level, and pedestrians use an artificial level one floor up.
These pods have many robocar features. They use rubber tires and run on open, unmarked pavement, guiding themselves via odometry and sensing magnets embedded every 5 feet or so in the pavement. They also have laser sensors which see obstructions on the roadway and any pedestrians. They will stop for pedestrians, and even follow you if you walk ahead, maintaining a fixed distance. The system is not designed to mix with pedestrians, however, and the control software shuts down the relevant section of the track if passengers exit their vehicle outside a station.
The tracking is accurate enough that, as you can see, the tires have left black trails on the pavement by constantly running in the same place.
Photos and video can currently be found at the PRT Consulting site and this video shows it pulling out of a station. There is only one other video — I hope more will arrive soon.
The economy has scaled Masdar’s plans back greatly. The original plan called for a whole city done one floor up with a network of these proto-robocar PRT pods running underneath, and no traditional cars in the whole city.
Submitted by brad on Fri, 2010-11-19 01:32.
Today, I was challenged with the question of how well robocars would deal with deer crossing the road. There are 1.5 million collisions with deer in the USA every year, resulting in 200 deaths of people and of course many more deer. Many of the human injuries and crashes have come from trying to swerve to avoid the deer, and skidding instead during the panic.
At present there is no general purpose computer vision system that can just arbitrarily identify things — which is to say you can’t show it a camera view of anything and ask, “what is that?” CV is much better at looking for specific things, and a CV system that can determine if something is a deer is probably something we’re close to being able to make. However, I made a list of a number of the techniques that robots might have to do a better job of avoiding collisions with animals, and started investigating thoughts on one more, the “flying bumper” which I will detail below.
Spotting and avoiding the deer
- There are great techniques for spotting animal eyes using infrared light bouncing off the retinas. If you’ve seen a cheap flash photo with the “red eye” effect you know about this. An IR camera with a flash of IR light turns out to be great at spotting eyes and figureing out if they are looking at you, especially in darkness.
- A large number of deer collisions do take place at dusk or at night, both because deer move at these times and humans see badly during them. LIDAR works superbly in darkness, and can see 100m or more. On dry pavement, a car can come to a full stop from 80mph in 100m, if it reacts instantly. The robocar won’t identify a deer on the road instantly but it will do so quickly, and can thus brake to be quite slow by the time it travels 100m.
- Google’s full-map technique means the robocar will already have a complete LIDAR map of the road and terrain — every fencepost, every bush, every tree — and of course, the road. If there’s something big in the LIDAR scan at the side of the road that was not there before, the robocar will know it. If it’s moving and more detailed analysis with a zoom camera is done, the mystery object at the side of the road can be identified quickly. (Radar will also be able to tell if it’s a parked or disabled vehicle.)
- They are expensive today, but in time deep infrared cameras which show temperature will become cheap and appear in robocars. Useful for spotting pedestrians and tailpipes, they will also do a superb job on animals, even animals hiding behind bushes, particularly in the dark and cool times of deer mating season.
- Having spotted the deer, the robocar will never panic, the way humans often do.
- The robocar will know its physics well, and unlike the human, can probably plot a safe course around the deer that has no risk of skidding. If the ground is slick with leaves or rain, it will already have been going more slowly. The robocar can have a perfect understanding of the timings involved with swerving into the oncoming traffic lane if it is clear. The car can calculate the right speed (possibly even speeding up) where there will be room to safely swerve.
- If the oncoming traffic lane is not clear, but the oncoming car is also a robocar, it can talk to that car both to warn it and to make sure both cars have safe room to swerve into the oncoming lane.
- Areas with major deer problems put up laser sensors along the sides of the road, which detect if an animal crosses the beam and flash lights. A robocar could get data from such sensors to get more advanced warning of animal risks areas.
Getting the deer to move
There might be some options to get the deer to get out of the way. Deer sometimes freeze, a “deer in the headlights.” A robocar, however, does not need to have visible headlights! It may have them on for the comfort of the passengers who want to see where they are going and would find it spooky driving in the dark guided by invisible laser light, but those comfort lights can be turned off or dimmed during the deer encounter, something a human driver can’t do. This might help the deer to move. read more »
Submitted by brad on Mon, 2010-11-15 15:20.
Many people wonder whether robocars will just suffer the curse of regular cars, namely traffic congestion. They are concerned that while robocars might solve many problems of the automobile, in many cities there just isn’t room for more roads. Can robocars address the problems of congestion and capacity? What about combined with ITS (Intelligent Transportation Systems) efforts to make roads smarter for human driven cars?
I think the answer is quite positive, for a number of different reasons. I have added a new Robocar essay:
Traffic Congestion and Capacity with Robocars
In short, a wide variety of factors (promotion of small, single passenger cars, ability to reverse streets during rush-hour, elimination of accidents and irrational congestion-fostering behaviour, shorter headways, metering of road usage and load balancing of roads and several others) could amount to a severalfold increase in the capacity of our roads, with minimal congestion. If you add the ability to do convoys, the increase can be 5 to 10 fold. (About 20-fold in theory.) The use of on-demand pooling into buses over congested sections allows a theoretical (though unlikely) 100-fold increase in highway capacity.
While these theoretical limits are unlikely, the important lesson is that once most of the cars on the roads are robotic, we have more than enough road capacity to handle our current needs and needs well into the future. In general, overcapacity causes building, so in time we’ll start to use it up — and have much larger cities, if we wish them — but unlike today’s roads which add capacity until they collapse from congestion, advanced metering can assure that no road accepts more vehicles than it can handle without major risk of congestion collapse.
Even before most cars are robotic, various smart-road efforts will work to improve capacity and traffic flow. The appearance of robotic safety systems in human driven cars will also reduce accidents and congestion along the way. Free market economist Robin Hanson believes the ability of cities to grow much larger will be one of the biggest consequences of robocar capacity improvements.
Submitted by brad on Mon, 2010-11-08 16:43.
I’ve written before about solutions to “range anxiety” — the barrier to adoption of electric cars which derives from fear that the car will not have enough range and, once out of power, might take a very long time to recharge. It’s hard to compete with gasoline’s 3 minute fill-up and 300 mile ranges. Earlier I proposed an ability to quickly switch to a rental gasoline car if running out of range.
A company called EMAV has proposed a self-propelled battery trailer to solve this problem. While I am not sure how real the company is, the idea has value, particularly when it comes to robotics. As I have written, robocars can solve the “range anxiety” problem in several ways; mainly that robots don’t care about how convenient charging is, and people don’t worry about the range of a taxi beyond the current trip. But batteries are still an issue, even there.
The trailer proposal has the car hitch on the small trailer (which has room for cargo as well) and it provides the extra batteries you need when dong a long trip. The trailer is also motorized so it puts no load on the possibly small car that is “towing” it. EMAV imagines you might buy this, keep it charged, and only put it on when you need to do a long trip.
That could work, but presents a few problems. First of all, cars are much less nimble when they have a trailer on them. Backing up is much harder, and in fact novices will get completely stymied by it. You take an extra-long parking space if you can fit at all. There’s also extra drag.
We might solve the maneuvering problem a bit with a mildly robotic trailer that has a link to the car controls, making backups and turns more natural. This can be done either with steerable wheels on the trailer or just independent motor wheels which can be turned at different speeds. Such a trailer might be able to couple much more closely with the car, possibly going right on the tail so that it acts like an extension of the vehicle. This might also solve the parking problem.
Things could also be aided by making the couple and decouple very simple and easy. That’s a tall order because of safety issues, and the need for a high-current wire. The ideal would be an automatic decouple, so you could temporarily drop the trailer off somewhere if you needed to handle roads and parking where a trailer isn’t workable. Even better but harder would be an automatic recouple, obviously requiring some more sophisticated robotics in the trailer, and a fully safe coupling system.
With standardization, trailers like this could be left on lots all over a city. Anybody with a compatible electric car could, if they needed it, stop off at a convenient lot to grab a trailer. (The trailer would also be in a charging station, making automatic coupling even harder.) With the trailer grabbed there would be no range anxiety. The trailer could simply provide power, or it could go further and charge the car at high speed, allowing the trailer to be dropped off at another charging station an hour or so later. (While this sounds nice, battery chemistries may doom this plan, since you now are putting two batteries through heavy use cycles to get one unit of charge into the car, doubling the battery lifetime cost of the energy.)
While eventually trailers would need to get back to their base after one-way trips, there are lots of ways to encourage various drivers to do that. As long as the dropped
trailer is not entirely empty, you can offer drivers who take it back a ride without using their own battery, for example.
This approach might be better than the battery-swap stations planned by “A Better Place.” The Better Place battery swap is cool, but requires all cars that use it be designed around its one particular battery configuration, and that people not own their own batteries. The swap stations are expensive and land intensive, while trailer depots would require nothing but a little land and a charging station for the trailer. A special trailer hitch is a much smaller modification of a car, too.
(One variation of the “PRU” trailer has the trailer contain a diesel generator rather than a battery pack. This of course has the range of liquid fuel, and doesn’t even need a charging station where you drop it of. It’s not being particularly green when used in this fashion of course, a bit worse than a serial hybrid car. If the trailer is heavy enough it could physically push the car and not need an electrical connection to it, though people might get highly confused by steering in such situations.)
As a cheaper and more flexible version of battery swap, this approach could be good for robocars too. Robots, unlike people, will not feel too burdened by the issues of driving a vehicle with a trailer, especially if they can control the trailer’s motors or steering. Parking’s easier too, especially if they can do robotic docking and undocking. While I have written how important it is that people don’t care about the range of a taxi, the owner of a taxi cares about the duty cycle. If they robotic taxi has to spend too much of its time recharging, the return on investment is not nearly as quick. The trailer approach, like the battery swap approach, means downtime only for the batteries, not the vehicle. If the trailers are themselves simple robocars, they can move at low and safe speeds to come meet robocars that need them for a range boost. Even if not, they need not take up much space and they’re easy to scatter everywhere for quick access. Indeed, the car itself might always use a trailer and thus have only enough battery power within it to get from one trailer to the next.
Submitted by brad on Thu, 2010-10-28 18:52.
I’m at the Pod Car City conference, taking place today and tomorrow in San Jose for PRT developers and customers. Some news tidbits from the conference:
- There were interesting presentations with videos from the three main vendors of working systems: ULTra (Heathrow), 2GetThere (Masdar) and Vectus (Swedish test track and some more.) Some were new videos showing the systems in action at a level not seen before.
- Sebastian Thrun of the Google robocar team gave his first outside talk on that project, with some great videos (not released to public, unfortunately.) Quite impressive to see the vehicle handling all sorts of traffic, even deciding when to cross over the solid line in the middle when it’s clear to avoid getting too close to parked cars, just as human drivers do.
- Sadly, during the public session (before Thrun’s talk) when several audience members sent in questions about the Google cars, both the host from San Jose and the leaders of the 3 PRT companies all punted saying they knew little about them.
- In spite of that there was intense interest in Thrun’s talk, with lots of questions and not nearly as much negativity as is sometimes directed at robocars from the PRT community.
- All vendors punted on my question about the current cost of pods (which external estimates suggest is around $100K since they are made in small quantities.)
- Lots of action in Sweden. Soon, a city will be chosen for a trial PRT system, probably either Stockholm or Uppsala. Then, a company will be picked — most people think it will be Vectus.
- Vectus, which makes a rail-based PRT, will be installing a system in Suncheon City, South Korea, which will be a people mover into the wetlands park there. Vectus showed many films of how well their system handles bad weather, though they are the only ones to use rails.
- In Masdar, one of the biggest challenges has been the oppressive heat, and the power for air conditioning. To make the PRT work, stations must be close as people will simply not walk long distances outside when it’s 40 degrees and humid.
- Interesting note about the rationale that helped sell ULTra at Heathrow: The big advantage is the predictable time of a PRT trip, which normally involves a pod already waiting and a direct trip. Even if that trip is no faster than a parking shuttle, not knowing when the parking shuttle bus will come is a major negative for those going to flights.
- Ron Diridon of the California High Speed Rail board declares that HSR will be a complete failure if there isn’t something like PRT around the HSR stations to disperse people into the towns. He’s half right — HSR is likely to be a big failure, PRT or not, though the PRT would help.
- San Jose is doing intensive study of a PRT to serve the airport, the nearby Caltrain and Light Rail stations, along with parking lots, rental cars and a couple hotels. This might well be useful but still is just a parking shuttle mostly. Few people take Caltrain or light rail to the airport (in spite of the existing free bus) and I doubt a lot more will.
- At the same time, thanks to ULTra, San Jose and other towns are starting to accept PRT as something costing 10-15 million per mile. That’s a lot cheaper than light rail, and in the bay area, hugely cheaper than the 50-year old BART system which people think of as modern.
- Attended a session on lessons from air traffic management for pod management. Interesting stuff but I don’t think that useful for the problem. Planes get spaced by 5 miles and 2,000 feet. Cars and pods will be spaced by tens of feet.
- Attended another session on trying to model passenger loads. This session was much more concerned about surge loads in many markets, where a class might let out and suddenly 100 people are at the PRT station trying to use it, removing the no-wait benefit (and the associated high predictability benefit.) One thing Robocars will probably do better since they have no concept of stations and you can get as many cars into an area as you can fit on the road and take out via it. Planners predict that if PRT waits are long in a campus situation, people will walk instead, but you would never have to walk instead with a robocar — just walk away from the crowd to get one.
- Still too much “transit” oriented thinking in the PRT crowd, I think. In fact, many are hoping to pitch PRT as a feeder which will increase usage of other transit lines. I think transit will fade away in about 25 years.
Vislab completes their autonomous drive to Shanghai
The team from the Vislab autonomous challenge made it to Shanghai, and their cars are now in the Italian pavilion at the World’s Fair. Congratulations to them. They sent me a nice PDF press release. It details elements from their blog about how they almost got a ticket, gave up driving at night, blew through toll booths, picked up hitchhikers, and could not handle crazy drivers.
Submitted by brad on Tue, 2010-10-19 14:10.
There’s news about the world’s two small PRT (Personal Rapid Transit) projects, which are both fairly robocar like, in that they involve self-steered rubber tired vehicles.
The Heathrow ULTra parking shuttle, which has 3 stops, uses pods which travel an ordinary pavement guideway with small curbs on the sides which the vehicle reads to guide itself. This system has been delayed many times, but reports it is finally getting close to opening. ULTra reports their system is now being used in a trial by the employees at LHR Terminal 5, though not the general public. Their newsletter also details how this demonstration system has spurred a fair bit of interest in a number of places, particularly India and Silicon Valley.
ULTra pods require dedicated track and don’t have tools to avoid pedestrians. The initial system is quite small.
The news is worse for the PRT system of Masdar, near Abu Dhabi. Masdar was planned as an all-green, car-free city. Aside from a maglev and light rail, the plan was to use an unusual arabian architecture. The ground level would be used by the PRT cars and other city vehicles, but a second “main floor” level would be built above this with the pedestrian walkways and main entrances to all the buildings. Arab streets are designed to be narrow, and elevated streets for people are easier to make than those for trains. To take the PRT, one would go into the “basement” which was really at ground level. In these streets, cybercars from 2getthere are to run. These are also rubber tire vehicles, but they get their guidance by following magnetic markers embedded in the road. Many people like the magnet approach as it is easy to follow, reliable and fairly cheap to install, even in existing roads.
Unfortunately as reported in Arabian News and this press release the plan for a city full of this “undercroft” and PRT cars has been scaled back. In particular, further reports indicate the elevated pedestrian street model will only be built in a limited zone at and around the university. The PRT will have only a 2 passenger stops, 1.7km of road and only 8 vehicles for the public (plus freight and VIP vehicles and 3 freight stops.) Outside this area will be conventional ground-level streets with conventional transit. It’s not been stated if private cars will also roam the regular streets, since the goal was a car-free city.
The article does make me wonder if they are thinking about robocars for those streets. The 2getthere cybercars do include some basic obstacle avoidance, though they are not ready to go out with the public. But at some point it will be possible, and the Masdar transportation plan might be realized even without the undercroft approach, or combining it with regular shared streets.
Robocars might also make sense for the Heathrow parking lot shuttle. Today the PRT is taking people to two stations in a large outdoor parking lot, but navigating a parking lot is a fairly easy problem for robocars — the environment is roughly controlled, and while there are pedestrians and slow-moving cars going in and out of parking spaces, these are easy to reliably avoid with LIDAR and other sensors. A better parking shuttle might take you right to your car, and pick you up there too if it is able to see people waving, or they text their spot number to a special number.
Of course, even better would be if the car then took you not to the airport entrance but into the airport, right to security. And even better if it was able to pick you up right at your gate and drive you (secured) along the controlled roads of the airport, out the security gate and right to the parking lot, taxi stand and rental car facility.
This has been a busy week of robocar developments, sparked in part by Google’s announcement. However, the PRT developments are just a coincidence.
Submitted by brad on Thu, 2010-10-14 14:09.
The AutoNOMOS team at Freie Universität Berlin led by Raul Rojas has shown a demo of their robocar acting as what I have called a whistlecar. Their latest car, named “Made in Germany” performed an autonomous taxi pickup for the press which you can see played out in this video.
They’ve built a wireless interface to their car, and in the demo, a developer uses an iPad application to send his GPS coordinates to the car and command it to come to him at a hotel’s entrance. The vehicle (which has been given more detailed maps including a map of the parking lot it is waiting in and the driveway of the hotel) then moves entirely vacant along streets to stop at the hotel entrance. Using the developer interface, they are also able to watch the car on the map as it moves to the hotel, and play out diagnostics on the iPad.
The FU-B team has been quite fond of wireless interfaces. Earlier this year for fun they built an iPhone app to access the drive-by-wire controls of their car so you could steer the car from outside using the iPhone. This is fun, but also in some ways a step back since the car has the ability to drive itself. Remote controlled cars are unexciting in comparison.
What is a major milestone for AutoNOMOS is that they have the confidence to operate Made in Germany entirely vacant on quiet city streets around their university, with the human supervision done by somebody with a remote control outside the vehicle. I would presume the vehicle, if it loses wireless connectivity, stops and attempts to assume a safe state, at least for now. Rojas says they haven’t done this very often, doing most testing on closed courses.
The whistlecar vision is an important one, I believe, for several key reasons. First, it may be deployable before robocars are considered safe enough to carry people at a speed they would accept, and as such it’s one of the incremental steps on the roadmap. Secondly, it enables car delivery, car sharing and autonomous refueling/recharging/servicing. By delivering a shared car that is the right car for a particular trip, transportation can become an order of magnitude more efficient than it is today when everybody rides alone in a sedan or SUV no matter what the trip.
Submitted by brad on Mon, 2010-10-11 17:02.
This weekend’s announcement that Google had logged 140,000 miles of driving in traffic with their prototype robocars got lots of press, but it’s not the only news of teams making progress.
A team at TU Braunschweig in Germany has their own model which has been driving on ordinary city streets with human oversight. You can watch a video of the car in action though there is a lot of B-roll in that video, so seek ahead to 1:50 and particularly 3:20 for the inside view of the supervisor’s hands hovering just over the self-turning steering wheel. There is some information on Stadtpilot here, but we can see many similarities, including the use of the Velodyne 64 line LIDAR on the roof and a typical array of sensors, and more use of detailed maps.
The team at Vislab in Milan has completed most of their Milan to Shanghai autonomous car journey which I have been following. You can read their blog or watch video (sometimes live) of their trip. A lot of the blog has ended up being not about the autonomous challenges, but just the challenges of taking a fleet of very strange looking vehicles in a convoy across Eastern Europe and Asia. For example, they have trucks which can carry their robocars inside, and once decided it was simpler to cross a border into Hungary this way. However, they left driving the vehicles, and the exit officials got very concerned that there was no record of the robocars coming into the country. I presume it wasn’t hard to convince them they were not smuggling Hungarian robocars out.
read more »
Submitted by brad on Sat, 2010-10-09 13:14.
Just released in a New York Times article and sidebar about highways and video, Google has unveiled an internal robot car project that has attained a remarkable level of robotic driving sooner than I and many others had predicted. The project combined the talents of Sebastian Thrun, leader of the Stanley/Junior team that won the Darpa Desert Grand Chellenge and came a close 2nd in the urban challenge, and Christopher Urmson from the CMU team that won the urban challenge and did second in the desert, along with 15 other engineers.
Their remarkable new Prius-based vehicles have completed over 140,000 miles of human-overseen driving on ordinary highways and city streets, including stretches of up to 1,000 miles without the human overseer feeling any need to apply a safety correction. By having a human in the car ready to grab the wheel, and a 2nd person also monitoring systems on a computer screen, the robotic operation on city streets is generally appraised to be legal.
As an example of the human intervention, during the test ride with reporter John Markoff, the human controller took the wheel when a cyclist ran a red light in a “just in case” intervention. Later examination of the sensors showed the car had indeed seen the bicycle and would have been expected to avoid it had the human not taken over.
This legal ability to have supervised driving should help build lots of great test data for robotic cars. Developers can build tools to try to judge whether, when a human intervened, the robot would have done anything particularly different, and look for those cases and judge and correct them. It also means, as I have described earlier, that we can start building the “trillion mile test suite” with all the data needed to do extensive virtual tests on new software revisions and prototype vehicles.
The new robotic Prius also looks a lot slicker than Junior, which very much has the experimental vehicle aesthetic. I expect the high resolution LIDARs to also get smaller and cheaper with time.
Later this week I should get a chance to see these vehicles up close and ride in one for more commentary. These results should make a stronger demonstration of how practical the technology is, to spur development and the legal steps necessary to move towards deployment when appropriate safety levels are reached.
Google of course is not a car company, but Sebastian Thrun has been involved there for some time as a creator of the street view camera car, and Larry Page has a longtime interest in transport innovation. Anthony Levandowski, creator of the Ghost Rider motorcycle entrant in the desert challenge and the PriBot (an earlier robotic Prius which he allowed to take him around the Bay Area while he supervised) is also a Google employee and on the team. Early research in robocars has come from academic labs and small teams, and it’s good to see Google get into funding groundbreaking work in the area.
Google is not a car company — but it has become one of the world’s leading companies in mapping an navigation.
In the long term, robocars should have a positive effect on society that exceeds even that of the search engine; this could become the biggest thing that Google does.
Submitted by brad on Sun, 2010-10-03 00:53.
There was a bit of a stir when Google last week announced that one of the winners of their 10^100 contest would be Shweeb, a pedal-powered monorail from New Zealand that has elements of PRT. Google will invest $1M in Shweeb to help them build a small system, and if it makes any money on the investment, that will go into transportation related charities.
While I had a preference that Google fund a virtual world for developing and racing robocars I have come to love a number of elements about Shweeb, though it’s not robocars and the PRT community seems to not think it’s PRT. I think it is PRT, in that it’s personal, public and, according to the company, relatively rapid through the use of offline stations and non-stop point to point trips. PRT is an idea from the sixties that makes sense but has tried for almost 50 years to get transit planners to believe in it and build it. A micro-PRT has opened as a Heathrow parking shuttle, but in general transit administrators simply aren’t early adopters. They don’t innovate.
What impresses me about Shweeb is its tremendous simplicity. While it’s unlikely to replace our cars or transit systems, it is simple enough that it can actually be built. Once built, it can serve as a testbed for many of PRT’s concepts, and go through incremental improvements. read more »
Submitted by brad on Tue, 2010-09-28 10:56.
An Italian team has built a prototype robot they call Dustbot which is aimed (in a backwards way) at the deliverbot vision.
The goal of the dustbot is to travel on demand to houses through the narrow, pedestrian streets of European cities so people can give the robot their trash, which it then takes back to the dump and drops there. It does not automate the pickup of the trash — you have to be there and put your bag into it, though it is able to drop it on it own. It is not clear if they plan to have it operate on streets with cars, or if it is truly ready to wander with civilians.
This is an evolutionary extension of the already common delivery robots used in factory floors and in hospitals. The hospital robots interact with the general public, and do it simply by being so slow that impact or injury is very unlikely, even with a programming error. But looking at the market of the very narrow, mostly or all-pedestrian ancient urban street, the challenge is more difficult than a hospital, but not as difficult as a vehicle that has to go fast enough that it could hurt somebody.
In tune with my predictions about deliverbots, the key is that the robot does not have to be in a hurry, so it can go as slow as is necessary to be safe. As the system improves, that speed gets faster and faster until it’s practical to go on urban streets at 15mph (ducking out of the way of cars) and eventually at the same speed as the cars. This robot can also be limited to a specific area in which it is well tested and armed with accurate data, because that’s much less of a restriction on delivery robots than it is on cars. (If you need to deliver elsewhere, use another service — but people will resist a taxi that will only take them certain places.)
Dustbot is probably too slow right now to be economical, particularly because you must wait for it. A robot that can pick up a standardized container is not too hard, however. One nice advantage of working on the trash problem is that there is no issue in leaving it on the street, so you don’t need to arrange home access for deliveries as a deliverbot would. There’s also little risk of piracy of the cargo, or damaging it.
There’s lots of video and photos on the site, here is a fluffy BBC video about the Dustbot. Note that this is about a year old — I just had not heard of it until recently.
Submitted by brad on Tue, 2010-09-21 10:14.
Here’s an idea that seems a bit wild and scary at first, but it’s doable today and has broad benefits: Small aircraft that don’t have landing gear, but instead land and take off from robotic “can’t miss” platforms pulled by cables on short airfields.
For every small aircraft purchaser, a big decision is whether to get retractable landing gear. They are very expensive, and create a risk of failure, but your plane will fly a lot faster and be more fuel efficient if you get them. What if we could leave the landing gear on the ground?
Imagine a wheeled platform on the runway with robotic control and a variety of systems to perfectly track an approaching aircraft. Pulled by cables, it can accelerate at several “g”s forward and back and left and right. As the aircraft approaches it tracks it and the cockpit display indicates positive lock. If the plane veers left, it veers left. If the plane speeds up it speeds up. Pretty much no matter what the pilot or winds do (other than missing the runway entirely) the plane can’t miss landing on it. It’s spring loaded so even if the landing is a bit hard the shock is cushioned. Done right, it’s just like having fancy shock absorbing landing gear. read more »
Submitted by brad on Sat, 2010-09-18 12:34.
This story from the Register about a test at the Stanford VAIL Lab reports an interesting result. They created a fake robocar, with a human driver hidden in the back. The test subjects then were told they could push the autopilot button and use the car. And they did, immediately picking up their newspapers to read as they would in a taxi (which is what they really were in.)
Not only that, when they were told the robot could not figure out the situation and needed human assist, they gave it, and then went right back to autopilot.
So trust of a robocar is already at a higher level than we might expect. I’ve ridden in Junior, and K. has stood in front of it, but that was with a human ready to take over the controls. Like many others pondering the future of robotic transportation, I believe we’ll only put robocars on our ordinary streets once they demonstrate a level of safety much superior to human drivers what I call the “robocar vision.” This does not mean a perfect level of safety, though, and the resulting accidents and occasional fatalities will be the cause of much debate and legal wrangling which will slow the development of the technology when it is saving lives.
Update: You might also like the Cute VW concept video where the dad explains to his son all the strange concepts like petrol, driving, traffic jams, accidents and parking.
Submitted by brad on Tue, 2010-07-27 00:50.
Today marks the start of a remarkable robocar trek from Italy to China. The team from the Vislab International Autonomous Challenge start in Italy and will trek all the way to Shanghai in electric autonomous vehicles, crossing borders, handling rough terrain and going over roads for which there are no maps in areas where there is no high-accuracy GPS.
This would be impossible today so they are solving that problem by having a lead car which drives mostly autonomously, but sometimes has the humans take over, particularly in areas where there are no maps. This vehicle can be seen by the other vehicles and also transmits GPS waypoints to them, so they can follow those waypoints and use their sensors to fill in the rest. The other vehicles also will have humans to correct them in case of error, and the amount of correction needed will be recorded. Some of the earliest robocar experiments in Germany used this approach, driving the highways with occasional human correction. (The DARPA grand challenges required empty vehicles on a closed course, and no human intervention, except the kill switch, was allowed.)
This should be a tremendous challenge with much learned along the way about what works and what doesn’t. As a computer vision lab, these cars appear to want to use vision a lot more than other robocars, which have gone LIDAR all the way. (There are LIDARs on the Vislab cars, but not as fancy as the 64 line Velodyne.)
They are using electric cars to send a green message. While I do believe that the robocars of the future will indeed be electric, and that self-recharge is a cruicial element of the value of robocars, I am not as fond of this decision. “One thing at a time” is the philosophy that makes sense, so I think it’s better to start with proven and easy to refuel gasoline cars and get the autonomy working, then improve what’s underneath. But this is a minor quibble about an exciting project.
They have a live tracking tool (not up yet) and a blog you can follow.
More robocar news to come. Yesterday I had an interesting ride in Junior (Darpa Grand Challenge II winner) and we trusted it enough to have Kathryn stand in the crosswalk while Junior drove up to it, then stopped and waited for her to walk out of it.
Submitted by brad on Thu, 2010-05-27 22:18.
I don’t often write about robots that don’t go on roads, but last night I stopped by Willow Garage, the robot startup created by my old friend Scott Hassan. Scott is investing in building open robotics platforms, and giving much of it out free to the world, because he thinks progress in robotics has been far too slow.
Last night they unveiled their beta PR2 robots and gave 11 of them to teams from 11 different schools and labs. Those institutions will be all trying to do something creative with the robots, just as a Berkeley team quickly made it able to fold towels a few months ago.
I must admit, as they marched out the 11 robots and had them do synchronous dance there was a moment (about 2 minutes 20 seconds in that video) when it reminded me of a scene from some techno thriller, where the evil overload unveils his new robots to an applauding crowd, and the robots then turn and kill all the humans. Fortunately this did not happen. The real world is very different, and these robots will do a lot of good. They have a lot of processing power, various nice sensors and 2 arms with 7 degrees of freedom. They run ROS, an open source robot operating system which now runs on many other robots.
I was interested because I have proposed that having an open simulator platform for robocars could also spur development from people without the budgets to build their own robocars (and crash them during testing.) A robocar test model is going to involve at least $150,000 today and will get damaged in development, and that’s beyond small developers. The PR2 beta models cost more than that, but Willow Garage’s donations will let these teams experiment in personal robotics.
Of course, it would be nice for robocars if there were an inexpensive robocar that teams could get and test. Right now though, everybody wants a sensor as nice as the $75,000 Velodyne LIDAR that powered most of the top competitors in the DARPA urban challenge, and you can’t get that cheaply yet — except perhaps in simulator.
Submitted by brad on Tue, 2010-05-18 15:37.
Last week, I attended a talk by Marc Raibert the former MIT Professor who founded Boston Dynamics, the makers of the BigDog 4-legged walking robot. If you haven’t seen the various videos of BigDog you should watch them immediately, as this is some of the most interesting work in robotics today.
Walking pack robots like BigDog have a number of obvious applications, but at present they are rather inefficient. BigDog is powered by a a 2 stroke compressor that drives hydraulics. That works well because the legs don’t need engines but can exert a lot of force. However, its efficiency is in the range of 2 gallons per mile, though this is just a prototype level. It is more efficient on flat terrain and pavement, but of course wheels are vastly more efficient there. As efficient as animals are, wheeled vehicles are better if you don’t make them heavy as tanks and SUVs.
BigDog walks autonomously but today is steered by a human, or in newer versions, can follow a human walking down a trail, walking where she walked. In the future they want to make an autonomous delivery robot that can be told to take supplies to troops in the field, or carry home a wounded soldier.
I wondered if BigDog isn’t trying too hard to be a mule, carrying all the weight up high. This makes it harder for it to do its job. If it could just tow a sledge (perhaps a container with a round teflon bottom with some low profile or retractable wheels) it might be able to haul more weight. Particularly because it could pay out line while negotiating something particularly tricky and then once stable again, reel in the line. This would not work if you had to go through boulders that might catch the trailer but for many forms of terrain it would be fine. Indeed, Boston Dynamics wants to see if this can work. On the other hand, they did not accept my suggestion that they put red dye in the hydraulic fluid so that it spurts red blood if damaged or shot.
The hydraulic design of BigDog made me wonder about applications to robocars. In particular, it seems as though it will be possible to build a light robocar that has legs folded up under the chassis. When the robocar got to the edge of the road, it could put down the legs and be able to climb stairs, go over curbs, and even go down dirt paths and rough terrain. At least a lightweight single person robocar or deliverbot might do this. read more »
Submitted by brad on Sat, 2010-05-15 14:14.
At the positive end of my prediction that robocars will enable people to travel in “the right vehicle for the trip” and given that most trips are short urban ones, it follows that most robocars, if we are efficient, will be small light vehicles meant for 1-2 people, with a lesser number of larger ones for 4-5 people. 2 person cars can even be face to face, allowing them to be under 5’ wide, though larger ones will be as wide as today’s cars, with some number as big as vans, RVs and buses.
Small, lightweight vehicles are not just greener than transit, they also require far less expensive road. While the initial attraction of robocars is that they can provide private, automated, efficient transportation without any new infrastructure, eventually we will begin building new development with robocars in mind. Various estimates I have seen for multi-use paths suitable for people, bikes and golf carts range around $100K to $200K per mile, though I have heard of projects which, thanks to the wonders of government contracting, soar up to $1M per mile. On the other hand, typical urban streets cost $2M to $3M per mile, an order of magnitude more.
Consider a residential robocar block. It might well be served by a single 10’ lightweight use lane. That lane might run along the backs of the houses — such back alley approaches are found in a number of cities, and people love them since the garage (if there is one) does not dominate the front of your home. It might also be in the front of the house. New construction could go either way. Existing areas might decide to reclaim their street into a block park or more land for the homeowners, with a robocar street, sidewalk and bike path where the road used to be.
We only need a single lane in one direction on most streets, though the desire to get 8’ wide vehicles in means there would be 2 lanes for the narrow vehicles. The lane would have no specific direction, rather it would be controlled by a local computer, which would tell incoming vehicles from which direction to enter the lane and command waiting vehicles to get out of the way. Small wider spots or other temporary holding spots would readily allow cars to pass through even if another vehicle is doing something.
You would not need a garage for your robocar as you can store it anywhere nearby that you can find space, or hire it out when you don’t need it. You might not even own any robocar, in which case you certainly don’t need a garage to store one. However, you probably will want a “delivery room,” which is something like a garage which has a driveway up to it. Deliverbots could use this room — they would be given the code to open the door — to drop off deliveries for you in a protected place. You could also have the “room of requirement” I describe in the deliverbots page.
This plan leaves out one important thing — heavy vehicles. We still need occasional heavy vehicles. They will deliver large and heavy items to our houses, ranging from hot tubs to grand pianos. But even heavier are the construction machines used in home construction and renovation, ranging from cranes to earth movers. How can they come in, when their weight would tear up a light-duty road?
The answer is, not surprisingly, in robotics. The heavy trucks, driven by robots, will be able to place their tires quiet precisely. We can engineer our robocar paths to include two heavy duty strips with deeper foundations and stronger asphalt, able to take the load.
Alternately, since the tires of the trucks will be further apart than our robocars, they might just run their tires on either side of a more narrow path, essentially on the shoulders of the path. These shoulders could be made not from heavy duty materials, but from cheap ones, like gravel or dirt. The trucks would move only very slowly on these residential blocks. If they did disturb things there, repair would be easy, and in fact it’s not too much of a stretch to predict either a road repair robot or a small road repair truck with a construction worker which moves in when problems are detected.
The volume of heavy trucks can be controlled, and their frequency. Their use can be avoided in most cases in times when the pavement is more fragile, such as when the ground is soaked or freezing. If they do damage the road, repair can be done swiftly — but in fact robocars can also be programmed to both go slowly in such alleys (as they already would) and avoid any potholes until the gravel robot fills them. Robocars will be laser scanning the road surface ahead of them at all times to avoid such things in other areas.
I keep coming up with dramatic savings that robocars offer, and the numbers, already in the trillions of dollars and gigatons of CO2 seem amazing, but this is another one. Urban “local roads” are 15% of all U.S. road mileage, and rural local roads are 54%. (There are just over 2.6 million paved road-miles in the USA.) To add to the value, road construction and asphalt are major greenhouse gas sources.
To extend this further, I speculate on what might happen if small robocars had legs, like BigDog.
Submitted by brad on Fri, 2010-05-14 23:59.
Last week, Volvo was demoing some new collision avoidance features in their S60. I’ve talked about the S60 before, as it surprised me putting pedestrian detection into a car before I expected it to happen. Unfortunately in an extreme case of demo disease known to all computer people, somebody has made an error with the battery, and in front of a crowd of press, the car smashed into the truck it was supposed to avoid. The wired article links to a video.
Poor Volvo, having this happen in front of all the press. Of course, their system is meant to be used in human driven cars, warning the driver and braking if the driver fails to act — not in a self-driving vehicle. And they say that had their been a driver there would have been an indication that the system was not operating.
While this mistake is the result of a lack of maturity in the technology, it is important to realize that as robocars are developed there will be crashes, and some of the crashes will hurt people and a few will quite probably kill people. It’s a mistake to assume this won’t happen, or not to plan for it. The public can be very harsh. Toyota’s problems with their car controllers (if that’s where the problems are — Toyota claims they are not — have been a subject of ridicule for what was (and probably still is) one of the world’s most respected brands. The public asks, if programmers can’t program simple parts of today’s cars, can they program one that does all the driving?
There are two answers to that. First of all, they can and do program computerized parts of today’s cars all the time and by and large have perfect safety records.
But secondly, no they can’t make a complete driving system perfectly safe, certainly not at first. It is a complex problem and we’ll wait a long time before the accident rate is zero. And while we wait, human drivers will kill millions.
Our modern society has always had a tough time with that trade-off. Of late we’ve been coming to demand perfect safety, though it is impossible. Few new products are allowed out if it is known that they will have any death rate due to their own flaws. Even if those flaws are not known in the specific, but are known to be highly likely to exist in some fashion. American juries, faced with minutes of a meeting where the company decided to “release the product, even though predictions show that bugs will kill X people” will punish the company nastily, even though the alternative was “don’t release and have human drivers kill 10X people.” The 9X who were saved will not be in the courtroom. This is one reason robocars may arise outside the USA first.
Of course, there might be cases the other way. A drunk who kills somebody when he could have taken a robocar might get a stiffer punishment. A corporation that had its employees drive when robotic systems were clearly superior might find a nasty judgement — but that would require that it was OK to have the cars on the road in the first place.
But however this plays out, developers must expect there will be bugs, an bugs with dire consequences. Nobody will want those bugs, and all the injuries will be tragic, but so is being too cautious on deployment. Can the USA figure a way to make that happen?
Submitted by brad on Mon, 2010-04-26 14:52.
This weekend I attended the annual “Robogames” competition, which took place here in the Bay Area. Robogames is mostly a robot battle competition, with a focus on heavily armed radio-controlled robots fighting in a protected arena. For several years robot fighting was big enough to rate some cable TV shows dedicated to it. The fighting is a lot of fun, but almost entirely devoid of automation — in fact efforts to use automation in battle robots have mostly been a failure.
The RC battles are fierce and violent, and today one of the weapons of choice is something heavy that spins at very high speed so that it builds up a lot of angular momentum and kinetic energy, to transfer into the enemy. People like to see robots flying through the air and losing parts to flying sparks. (I suspect this need to make robots very robust against attack makes putting sensors on the robots for automation difficult, as many weapons would quickly destroy a lot of popular sensors types.)
The games also featured a limited amount of automated robot competition. This included some lightweight (3lb and 1lb) automated battles which I did not get to watch, and some some hobby robot competitions for maze-running, line following, ribbon climbing and LEGO mindstorms. There was also semi-autonomous robot battle called “kung fu” where humanoid robots who take high level commands (like punch, and step) try to push one another over. There is also sumo, a game where robots must push the other robot out of the ring.
I had hoped the highlight would be the Robo-magellan contest. This is a hobbyist robot car competition, usually done with small robots 1 to 2 feet in length. Because it is hobbyists, and often students, the budgets are very small, and the contest is very simple. Robots must make it through a simple outdoor course to touch an orange cone about 100 yards away. They want to do this in the shortest time, but for extra points they can touch bonus cones along the way. Contestants are given GPS coordinates for the target cones. They get three tries. In this particular contest, to make it even easier, contestants were allowed to walk the course and create some extra GPS waypoints for their robots.
These extra waypoints should have made it possible to do the job with just a GPS and camera, but the hobbyists in this competition were mostly novices, and no robot reached the final cone. The winner got within 40 feet on their last run, but no performance was even remotely impressive. This was unlike past years, where I was told that 6 or more robots would reach the target and there would be real competition. This year’s poor showing was blamed on budgets, and the fact that old teams who had done well had moved on from the sport. Only 5 teams showed up.
The robots were poor for sensors. While all would have a GPS, in 1 or 2 cases the GPS systems failed and the robots quickly wandered into things. A few had sonar or touch-bars for obstacle detection, but others did not, and none of them did their obstacle detection well at all. For most, if they ran into something, that was it for that race. Some used a compass or accelerometers to help judge when to turn and where to aim, since a GPS is not very good as a compass. read more »