The bi-annual Conference on Automated People Movers and Automated Transit Systems was held this year in Phoenix, Arizona. While this conference has traditionally been targeted at builders and operators of large Automated People Movers, there seems to be a growing realization within this community that the next step in the evolution of automated transit is to smaller vehicles that provide point-to-point, on-demand service. With that in mind, here is a recap of some of the conference highlights.
The keynote address at the opening Plenary Session was delivered by Ed Finn, the director of The Center for Science and the Imagination at Arizona State University (csi.asu.edu). Dr. Finn began by recalling a 1940 short story The Roads Must Roll in which Robert Heinlein foresaw the dominance of the automobile, as well as the limitations of oil as a strategic resource. The system of automated mechanical roadways he foresaw enabled people to live in the country and work in the city – the conceptual beginning of the modern suburb. But the story is not so much about a technology as it is about the effect that that technology could have on culture. This story, from the golden age of science fiction, illustrates that the first step in building the future is to imagine it. This is the mission of The Center for Science and the Imagination – to create a better future through thoughtful optimism.
The idea for the Center began with a talk by science fiction writer Neal Stephenson in which he complained that our vision of the future has become dystopian, and asked where are our big new ideas? The president of ASU, Michael Crow, was in the audience, and responded that it is the science fiction writers who should be helping us to dream better dreams. That got people thinking about how we can put science fiction back in conversation with science and engineering to change the way we think about the future. The mission of the center is to get people thinking more creatively and ambitiously about the future, to encourage fresh visions and radical experiments, to remind us that we are all invested in the future, and that the future is created by the choices that we make every day.
The Center is pursuing these goals in several ways. The Hieroglyph Project is fostering interdisciplinary collaboration between scientists, engineers, artists, and writers. For example, Neal Stephenson was paired with structural engineer Keith Hjelmstad. Stephenson asked, “How tall could we build something?” This is the sort of question that challenges us to explore new ideas. To keep things interesting, the project is focused on concepts that could become reality within one lifetime. The final product will be a published anthology of stories.
In another initiative, Dr. Finn will be teaching a course called “Prototyping Dreams” in which students will explore ways to communicate their vision to a broad public. They will do this by building tangible prototypes that, while not necessarily functional, can communicate an idea, and tell a story. It is hoped that these artifacts can empower us to suspend disbelief, fill in the details, and experience a possible future. And if that future is compelling enough we might actually decide to build it.
Dr. Finn observed that when we have resolved a technological problem, it essentially disappears, and he speculated about a future in which transportation becomes as invisible as electricity is today. He encouraged us to think of urban and transportation planning as a kind of design fiction – an effort to create a utopia. We need to imagine not just the places we want to go, but the people we want to be.
Brad Templeton, a director of the Electronic Frontier Foundation, and author of robocars.com, spoke on Where robot cars can really take us. (Mr. Templeton has done consulting work for Google but was not representing them.) The primary motivation for “smart car” research has been saving lives, but other goals include saving energy, reducing parking and traffic jams, and delivering goods. Human drivers are not very good – 33 thousand are killed every year in the US, 1.2 million worldwide. The cost of accidents works out to about 8¢/mile. Google has just logged 90,000 miles without a safety-critical incident. Mr. Templeton claimed that that’s an accident rate close to human beings. In the Q&A session, Martin Lowson of Ultra Global, Ltd., noted that human drivers go much further than that before an accident, so it is too soon to conclude anything from Google’s brief experience. In the UK there are about 2.27 traffic deaths per billion passenger kilometers traveled. [According to NHTSA, in the US in 2010 it was about 1.10 deaths per 100 million vehicle miles.]
The technology that has made this possible is the scanning laser rangefinder (LIDAR). A central premise of robocar research is that it’s not necessary to identify every obstacle – just avoid everything. It is hoped that this will not require the intelligence of a human being, but perhaps only that of a horse, or even an insect. The cars also have radar, which is less precise than LIDAR, but has the advantages that it can see through fog, and it can measure the speed of moving objects. There is also GPS, but that doesn’t always work, so you can’t rely on it.
Robocars offer the flexibility to order the right vehicle for each trip, a small one for one person or a large one for the whole family. They are also the best way to green transit. The average bus in the US has 9 people on board, and gets 4 to 5 MPG. That’s about the same energy use as the average car. But new lightweight energy efficient cars, such as Edison2, are far more efficient. Electric cars and alternative fuels will get a boost because charging and refueling stations do not need to be ubiquitous – autonomous vehicles can travel to where the fuel is. Robots don’t care about how convenient it is to refuel or recharge. They don’t care about anything – they’re robots. Because these vehicles can be shared, we will need vastly less parking. (There are at least 3 parking spaces for every car in the US. It has been estimated that 60% of the land of Los Angeles is devoted to cars.) He believes smart cars will increase the capacity of our existing roads. With the addition of delivery robots, we will be able to deliver any product to your door within 30 minutes. What will this mean to retailing, to manufacturing, to the very concept of ownership?
Importantly, all this can be done without any new infrastructure, which means the technology can be deployed quickly. Templeton is optimistic that we will see commercial self-driving cars within this decade. This will be driven by early adopters instead of municipal government administrators. Early adopters are great because they drive innovation, and he thinks there are enough of them to get us through the first phase of deployment. Templeton believes that a bottom-up approach, where innovation happens at the edges, is the best way to bring innovation to transportation.
What does this mean for APMs? No need for a guideway or a right-of-way, or stations. This will mean much lower costs. There are places where you need new infrastructure, but we should follow the model of the Internet, and make the infrastructure as simple as possible. Bare concrete is very simple, and it can accommodate cars, buses, bicycles, pedestrians, or just about anything. Infrastructure that is designed for a specific vehicle lacks flexibility. Many competing types of vehicle can deliver better results. [If it can accommodate buses or “just about anything”, then it must be extremely robust, which is prohibitively expensive. In addition, while versatility is great, pedestrians and cyclists don’t particularly want to share their space with buses and trucks. Any cyclist will tell you that they prefer to ride on a dedicated bike trail.]
At the Roundtable on the State of the APM Industry, much of the discussion continued to revolve around the relative merits of robocars versus guideway-based transit. Martin Lowson pointed out that Google has run 500,000 miles, but they have a one person at the wheel and another at the computer. The Ultra PRT (Personal Rapid Transit) system at Heathrow has already run 1 million miles with only passengers in the cab. 2GetThere have done several hundred thousand as well. Templeton countered that an inferior technology with a faster pace of innovation is always the better bet. He conceded however that it’s amazing how willing people are to trust the Google car, “They trust it more than they should.” Mr. Lowson’s vision of the future of PRT is to gradually relax the constraints so they can drive in more complex environment, eventually becoming dual-mode. He believes that that is a quicker and more practical approach to introducing autonomous vehicles.
The panel discussion on Autonomous Road Vehicles touched on a wide range of issues including technology, law, culture, economics, regulation, urban planning, environment, insurance, public acceptance, accessibility, freight, retailing, parking, and ride sharing. Brad Templeton showed a video (http://youtu.be/cdgQpa1pUUE) documenting the one-week loan of a Google car to Steve Mahan, CEO of the Santa Clara Valley Blind Center. Mahan, who is “well past legally blind” said “This would change my life.”
Gary Marchant Professor of Law at Arizona State University said that he has noticed that some people are “exuberantly overexcited” about driverless cars, and others are very skeptical. Larry Page is on record as a huge fan of PRT. Templeton used to be, but now feels that self-driving cars are more promising. He noted that, from the opening of the Morgantown PRT in 1979, it took until 2006 to get the first sale of a PRT system at Heathrow. He decided that the problem is that when you have to get approval from a wide variety of stakeholders, innovation moves at a very slow pace. [He neglected to mention however, that the first sale of a self-driving car has yet to occur.]
Both Mr. Marchant and Brad Smith, Fellow at Stanford University agreed that it is likely that legal impediments will delay the commercialization of autonomous cars. Templeton pointed out that there are hundreds of thousands of car accidents every day, so we already have a mechanism for paying for them. We could continue to use that model, or a similar one, for self-driving cars. Some of the cost may be shifted from the insurance companies to the car companies, but it’s all ultimately paid for by the consumer. Martin Lowson noted that that’s assuming that you have enough data to make statistical conclusions. Current experience with the Google car is nowhere near that, and there’s no way to get enough data within any reasonable time scale. Templeton suggested that one possible solution for this might be to measure the frequency of near accidents, and use that to estimate the frequency of actual accidents. Bryant Smith observed that the gradual introduction of automated features into automobiles actually presents a greater legal challenge than full automation because it’s less clear who is liable. Gary Marchant noted that judgements against car companies have ranged from $5,000 to $4,000,000,000. Where self-driving cars end up on that scale will make a huge difference. And having the technology that performs best overall will not protect you from lawsuits. Since cases are judged individually, if your technology is not as good as someone else’s in that scenario, even if your’s is better overall, you will be liable. Bryant Smith pointed out that there are three kinds of uncertainty: psychological, legal, and reputational. Bad press could be far more damaging to a car company than a large judgement if it causes a drop in sales. Templeton agreed that if the deep pockets of car companies cause the size of judgements to soar, that could kill the smart car.
Templeton thinks that some people will always use their private robocar, some will always use a robocar taxi service, and some will own a private robocar, but will sometimes rent a different car for special purposes. He predicted that cars with Vehicle-to-Vehicle (V2V) or Vehicle-to-Infrastructure (V2I) communications, so-called “connected cars”, are not likely to succeed, in part because the first cars that have the feature will not have anything to talk to. LIDARs don’t interfere with each other, but there is a potential for interference between radars.
Someone from the audience asked how many robocars Google has. Templeton replied that they haven’t said, but if you stand on the road near Google headquarters, you’ll see one go by every few minutes. So it would seem that there are at least a dozen.
Mr. Templeton thinks that self-driving cars will eliminate the need for dedicated guideways and rails, and that there isn’t enough traffic volume in most places to justify a grade-separated guideway. Martin Lowson observed that everyone is assuming that the future will combine autonomous and manual vehicles, but it’s massively easier to develop autonomous vehicles that only have to interact with other autonomous vehicles. Multi-purpose roadways must be able to carry 44 ton trucks, but 90% of the vehicles are much lighter, so our infrastructure must be massively overbuilt. Putting computers into cars will not change this. But Templeton believes that people want a vehicle that an go anywhere. The first law of robocars is don’t change the infrastructure. You can solve the problem technologically while others are debating how to change the infrastructure. [Mr. Templeton apparently sees guideways as a large expense that is made necessary by the limitations of PRT vehicles. Our own view is that guideways are an inherent advantage that is made possible by the characteristics of PRT vehicles. It is certainly impressive that the Google car is capable of stopping for traffic lights, stop signs, pedestrians, and other vehicles. But it is much more impressive that PRT pods don’t need to stop for any of these things. The guideways also create the opportunity for superior architectural designs and land-use planning.]
After 20 years of work, the new APM Standard is completed, and the 51 (unpaid) members of the committee are very proud of it. It complies with ANSI and ASCE standards. Larry Smith, the chairman of the committee, said they desperately wanted to have the printed document available for sale at the conference, but the publication of both the print and electronic versions has been delayed until August. He was actively soliciting new committee members for the next revision. [I attended a meeting with some members of the committee in September 2007 at the offices of Taxi2000 near Minneapolis. At that meeting many people urged the committee to make the standard broad enough to be applicable to PRT, and the new standard does now allow for off-line stations. Hopefully the updated standard will help to provide a regulatory framework for new ATN (Automated Transit Network) deployments.]
In PRT at Heathrow: Delivering World Beating Customer Service, Martin Lowson of Ultra Global, Ltd., gave an update on the performance of their system since its opening in April 2011. Heathrow regularly use passenger surveys to evaluate performance of services throughout the airport. Their results show that the Ultra PRT system rates as the best single service in the entire airport. 94% of all passengers surveyed perceived PRT to be better than the shuttle bus service it replaced, the remaining 6% said they were equal. No one preferred the bus. Comments on Twitter have also been overwhelmingly positive. Use of the business car park at Terminal 5 is up enormously despite a 23% price increase. BAA now plan to expand the capacity of that car park, and have announced that expansion of the pod system into other parts of the airport is part of their future planing. Ultra has exceeded all their specification for performance and reliability.
In Masdar Experience, Robbert Lohmann of 2getthere reviewed the first two years of operations of their PRT system in Abu Dhabi. They have been getting three times the number of passengers than were originally expected, and on November 20, 2012 they carried their 500,000th passenger. Since the last control system upgrade, system availability has been stable at 99.5%. One of the longest service interruptions they have had occurred shortly after the system opened. A passenger in a temporarily stopped pod, pushed the emergency button, and got out of the pod. This caused another vehicle in the vicinity to stop, so he pushed the emergency button on the outside of that vehicle, letting those passengers out onto the track as well. Mr. Lohmann used a YouTube video from a fun-loving tourist (http://youtu.be/_0nwOq5GMsQ) to illustrate how a person who is unfamiliar with the system knows intuitively how to use it. [You can also see how much fun people have using it. We engineers sometimes lose sight of the fact that people expect more from transportation than transportation.] 2getthere has a total of 9 people in Masdar: an operations manager, a chief service engineer, 2 service engineers, and 5 operators. (Station security and cleaning are handled by other Masdar personnel.) During the first year, there were always two operators in the control room, but now it is considered adequate to have only one operator, provided a second person is available in the maintenance facility. This allows the control room to remain staffed even if it is necessary for someone to go out to the system. Mr. Lohmann gave an overview of their Transit Operations Monitoring & Supervision (TOMS) system. Due to the extremely hot climate, the vehicle chargers in the terminals are used only to power the pod’s air conditioners, not to charge the batteries. Pods can go up to 60 kilometers, or 5 hours of continuous operation, on a charge. After that, they are sent to a charging room, where they reach full charge in 1½ to 2 hours. There are video cameras in the pods and stations, but not on the track. The flatbed vehicles that 2getthere designed have not yet been put into service because it has not been determined how to get freight to and from the freight stations.
The original design for Masdar was built, literally and figuratively, on an undercroft – a level below the pedestrian level where the PRT system could operate. This created the opportunity to revive the traditional Arabic city plan with its narrow pedestrian streets, where shade and continuous breezes keep the outdoors cooler, or at least less hot. But with the recent economic downturn, the developers are now reconsidering their goal of creating a demonstration to the world, and looking for ways for the project to earn a profit. In an effort to lower costs, they are trying to eliminate the undercroft, but it’s not clear if there’s a way to do that while preserving the comfort and energy efficiency that have attracted worldwide attention to the project.
In San José International Airport Automated Transit Network Feasibility Study, Gary Hsueh of Arup presented a summary of their recently completed study. He said that although the City has not identified funding, they are still interested, and are presently gathering public agencies to discuss how innovation can happen in the public realm. As mentioned by the city’s Director of Transportation in his presentation to the Transportation Committee of the San José City Council on November 5, 2012, they are now looking at a more compelling application serving Diridon Station, providing better access to the planned high-speed train service. In answer to a question from Martin Lowson, Mr. Hsuch said that their estimated cost for the San José guideway was about five times the actual cost of the guideway at Heathrow. He attributed this to the more stringent seismic requirements in California, the addition of evacuation walkways along the entire length of the guideway, the span length, the choice of materials, the assumption of a much heavier vehicle, and a generous “risk contingency”. He explained that the high risk contingency was applied even to well-understood parts of the project partly because of the uncertain regulatory environment. It is still unclear whether regulatory agencies would accept the APM standard as relevant or sufficient. Arup’s recommended path forward would be for the city to refine the design in partnership with a “preferred supplier” as a way to reduce some of the contingencies.
One consultant at a firm that does not specialize in ATNs told me that he had only skimmed the studies and read the executive summaries, but confided that he was “disappointed but not surprised” by the findings. At a reception at the Mesa Arts Center, I had the opportunity to chat with an official from the City of San José. I mentioned that some of us were concerned when the studies were deleted from the city’s website, along with the page proclaiming the city’s interest in ATNs. She assured me that they have not lost interest in ATNs, and that the missing materials were only temporarily unavailable during a website reorganization. And indeed they can now be found under Transportation>Green Mobility>Automated Transit Network. In explaining their caution, she mentioned the finding of the Aerospace report that ATNs are “orders of magnitude more complex” that other forms of transit. This is an issue that warrants further discussion. Perhaps it will receive some attention at the upcoming meeting in June at the offices of Aerospace Corp.
Robbert Lohmann of 2getthere reported on Station Dynamics Influencing Ride Sharing. He introduced his topic with a series of animated commercials for a Belgian transit company called It’s smarter to travel in groups (http://youtu.be/llE8FtJnf-k). Most people are willing to ride-share, particularly at peak travel times. (Mr. Lohmann noted that Islamic countries can be an exception. There, if a man tries to share a pod with traditionally dressed ladies, he can expect to be the only one left in the vehicle.) At Masdar, where there are only two passenger stations, the average vehicle occupancy is 2.4. But as the number of possible destinations increases, ride sharing becomes more difficult. Some of the approaches that were considered are:
- Give priority to passengers who are sharing.
- Implement a social media-based rewards program.
- Reserve berths closer to the station entrance for ride sharing. (The control screen in the pod would list a subset of all possible destinations.)
- Charge extra to ride alone.
It was ultimately decided that a combination of all of these would be most effective.
Mr. Lohmann also shared some of 2getthere’s innovative ideas for Sustainable APM Infrastructure at Airports. These included:
- Reusing old containers as the guideway.
- Using existing trees to support the guideway.
- Collecting and storing rainwater in the guideway to prevent it from freezing in winter.
- Bamboo for guide-rail – one of the hardest and most sustainable woods.
- Solar panels that are largely transparent, and can be walked on can be integrated into the guideway, and could generate up to 47% of the energy used by the system. They are also dirt repellant to maintain efficiency.
- Use of Fiberline composite road decking (fiberline.com) to reduce guideway weight, and maintenance costs. A coating is applied to increase traction.
Martin Pemberton of Vectus, Ltd. was unable to attend, so he asked Peter Muller of PRT Consulting to give their presentation, The Track to Suncheon: Making APMs Intelligent. (PRT Consulting is not affiliated with Vectus or any other PRT supplier.) For the Suncheon project they have made a number of changes from their original test track in Uppsala. In Uppsala they used a guideway with built-in linear motors, but no power rails. In Suncheon, the guideway has power rails (500 volts DC), and the pods have rotary motors. They have developed a clever means of mixing PRT and GRT (Group Rapid Transit) on the same guideway. The larger vehicle is made by combining two of the smaller vehicles with an articulated connecting section. The doors on the larger vehicle have the same spacing as those of two adjacent smaller vehicles, so a single station design with in-line berths can accommodate either size of vehicle. For Suncheon, they are using 40 of the smaller pods, each of which can seat six adults, plus another six standing. The pods have doors on both sides, and solid wheels that run on steel tracks. They use a distributed, asynchronous, dynamic moving-block control system. That is to say, the control computers are distributed throughout the network, rather than centralized at one location. Asynchronous control systems make vehicle-control decisions based on local conditions as a vehicle proceeds along its route, instead of pre-planning the entire trip. In a moving-block system, the protected area around a pod moves along with it, as opposed to a fixed-block in which the blocks are at fixed locations along the track. A dynamic block changes length depending on how fast the pod is moving. The pods will operate at up to 43 MPH with headways of three to four seconds. Pods are being delivered more slowly than expected, so the opening has been delayed to this fall. They anticipate about three million passengers per year.
Two presenters from the Warsaw University of Technology described their paper, Personal Rapid Transit – Computer Simulation Results and General Design Principles. Dr. Wlodzimierz Choromanski began by saying that he personally does not believe in high-capacity PRT. As he said this, he seemed to scan the crowd to gauge the reaction this would get. He went on to claim that he can “prove” that it is “impossible” to combine the features of individual transport with mass-group transport, and that an effort to do so would only move traffic jams from ground-level to overhead. [By high-capacity, does he mean transporting a small number of large groups between a small number of locations, or a large number of small groups between a large number of locations? The latter is the far more common need, and PRT does that better than any other form of public transit – perhaps better than cars.] But they do feel that PRT is a good fit for many specialized applications. Due to the aging of the European population, they placed particular emphasis on accommodating people with disabilities. Dr. Bartlomiej Kaminski then took the stage to explain their design for a linear induction motor powered by inductive power transfer. There are well known advantages to linear motors – immunity to loss of traction (for safe operation even in ice or rain) and low maintenance (because there are no moving parts). But they are still developing control algorithms that can adjust for the deterioration of thrust force with speed, and with the size of the air gap. Their design is being tested on a heavily instrumented, reduced-scale test track. The advantages of wireless power transfer include low maintenance because there are no power lifters to replace, and greater safety due to the absence of high-voltage power rails. They have been able to transfer 1.2 KW on the test track with 90% efficiency.
In the second presentation from Warsaw University of Technology, PRT (Personal Rapid Transit) Computer Network Simulation, Analysis of Flow Capacity, Wiktor Daszczuk, described how they started with existing simulation tools, but ultimately decided to write their own so they could use their own algorithms, and get detailed output so they could examine individual events and track vehicles. Ultimately they wrote two simulators, one using cellular automata and the other event-based. The cellular automata simulator is extremely fast, but not quite as accurate at the event-driven one. They found that running hundreds of simulations with a range of parameters was a tedious and error prone process, so they created an automated front end that generates the models and stores results in an SQL database for easy analysis. To test performance under saturation conditions, they load stations with an infinite number of passengers.
In casual discussions at meals or coffee breaks, I sensed a growing realization that there are not likely to be many more opportunities in the future to build multi-billion dollar APM projects. This may explain the growing interest in ATNs as a potentially larger, but lower cost, market. I chatted with a guy from Bombardier who confided that they have periodically reexamined the idea of entering the PRT field, but have always been deterred by the R&D expenditures. One representative of a cable-car company stressed that they had a unique advantage in being able to deliver systems quickly. In particular, their ability to delver a system within one election cycle had often proven decisive.
I last attended this conference in 2009 when it was known as the Automated People Mover Conference. There has been a marked change of attitude towards Automated Transit Networks since that time. So much so that the conference is now called Automated People Movers and Advanced Transit Systems. Even the logo have been updated to reflect the change. In 2009 there were no sessions specifically devoted to Personal Rapid Transit. In 2013 there were three. And the newly completed revision of the APM standard now specifically allows for off-line stations. It would seem that years of persistent effort by ATN innovators are beginning to pay off.
While I was finishing up editing this document, I learned of the passing of Martin Lowson. I have not altered the content since that time, so if it seems that Martin played a significant role in the conference, it’s because he did. Fittingly, his presentation on Delivering World Beating Customer Service was something of a victory lap for him. And while many attendees seemed to be thinking for the first time about the ramifications of autonomous vehicles, Martin was well prepared (as always) to explain the careful analysis and sound reasoning that led to the design of the Ultra PRT system. Perhaps it is a cliché to say that he will be sorely missed, but he really will be.