TheTigerMaster
Superstar
The collision avoidance is already far better than what a human driver could ever dream of. Google's car has yet to be in a single at fault accident after hundreds of thousands of miles on the road.
|
|
|
Automated driving & Transit
- Thread starter ehlow
- Start date
The collision avoidance is already far better than what a human driver could ever dream of. Google's car has yet to be in a single at fault accident after hundreds of thousands of miles on the road.
diminutive
Active Member
Apparently google has taken their cars through some pretty dense areas of San Francisco, so presumably they've been dealing with lots of the complexities you mention. In any complex situation though, it's likely an automated system will have better awareness and safer behaviour. Human drivers, today, routinely screw up driving around streetcars.
Even without fully automated driving though, partial automation could conceivably improve road safety and capacity quite substantially. For instance, a kind of lane-keeping system with adaptive cruise control and collision avoidance. Rather than having drivers manually turn one way or another, the system could respond to driver-given turn signals and execute the turn when safe, based on nearby vehicles.
Granted, the full on automatic-car utopia seems a bit Jetson-like, but there's clearly a lot of room for innovation in the car world.
Combined with improvements to conventional engines and battery-electric vehicles, which will likely reach cost parity soon as well, there really is quite a lot of innovation going on with automakers. There's a good chance that our roads will be substantially more efficient and our vehicles far less environmentally destructive.
ehlow
Senior Member
I'm guessing that when it's first implemented, the TTC will still have someone sitting in the driver's seat ready to take control, as well as handling the opening & closing of the doors. I'm also guessing adoption of the technology will be very slow, because the TTC is liable if the automated vehicle hits someone or something. I mean, I'm guessing they're liable now if a human driver does, but transferring that trust to software will likely take time and lots of testing.
By the way, how does operating the doors work on the Skytrain? Is someone controlling it looking at the station through a camera? Or is that automated using sensors?
rbt
Senior Member
In near perfect weather with clear visibility. Google does not take its cars out when roads are slipperly (rain, ice, snow), or visibility is low (fog, snow).
If you look at accident rates among humans you'll find they are also lowest during clear weather with good road conditions; though not as good as Google's record thus far.
More specifically, their system doesn't work when it can't see the lines on the road or a transition in material (pavement to gravel). It is not yet a replacement for drivers in all weather conditions that we would expect transit to be running in.
Last edited:
rbt
Senior Member
Automated sensors, similar to an elevator. It was a fun game for a while to put a cardboard box in the doors of an empty train and see how long it took transit control to send someone out to remove it.
Dragging passengers sometimes happens (doesn't hit the news for the same reason TTC keeps various deaths quiet) but that's not limited to just automated systems.
Last edited:
ehlow
Senior Member
rbt
Senior Member
Stations are monitored but it's not the primary mechanism for the doors.
Last edited:
georgevicbell
Active Member
Glad to see that this is being discussed, I was just about to create a thread myself...
Metrolinx and Government of Ontario need to start creating an environment where this can begin to be studied and developed in ontario. Leaving this to california, or germany is short sighted. There are a number of things that they should be doing in the short, medium and long term to encourage development of all three levels of automated driving which are -
1) warning and avoidance
2) controlled conditions (right of ways) - driver takes over when car is not sure
3) completely automated
Some of the things they should be doing:
1) Creating zones and corridors that are free to experimentation in the areas of car-to-car and car-to-infrastructure networking - for example, Waterloo in the area of the university - if people want to connect devices that communicate with vehicles to parking structures, street lights, light posts etc. there should be a very clear way of engaging the government to get that done. There are "fake" towns that are used by police services, driver training etc. that could be used in off hours for this testing as well. This will encourage research and development. It's also necessary to provide sections of highways, rural roads etc. Systems that solve the "warning and avoidance" item above could be intersections that communicate with cars to tell them if the light is red or green - and the car can then warn a user they may be about to drive through a red light. These systems should be fairly easy to implement and test.
2) All the transportation systems and boards should be required to provide open and free access to up to date data - on parking availability in Go Parking lots, on metered parking, etc. if the data doesn't exist or we can't get it, we should be challenging people to develop ways to get it and create it.
3) For number 2, we have a number of routes that are already protected (subways), and some routes that are partially protected (LRT/streetcars on spadina, lakeshore, st. clair, BRT's in vaughn etc.) there should be efforts made to make these routes as safe as possible using current technologies, but potentially to also upgrade them to fully automated where possible.
4) We also have a number of roads where lanes could be made partially protected or fully protected, or act in such a manner, I'm thinking areas of the 401 between windsor and waterloo, areas of the 400 north of parry sound->sudbury and beyond. I imagine that if the government offered a roadmap to self-driving freight on these routes that a number of companies would be willing to work on creating such a system. There could be a number of clear rules - trucks must be clearly labeled as self-driving, they must not operate in inclement weather, all systems must prove themselves with XXX,XXX of hours "attended" driving, they must pull-over if traffic becomes heavy, they must not be carrying hazardous material, they must not travel through cities automatically but park and be driven to their final destination. I would imagine that it would even pay for companies to outfit the road with censors to increase the speed they could deliver such a system (that check for incursions, or road/weather conditions). Work with fedex or a number of other companies that ship lots of stuff by truck.
5) We should be investigating systems that provide for less drivers of transit vehicles (bus trains etc.) that could be implemented without much difficulty in BRT conditions.
6) We should start to publish maps that show where infrastructure-to-car systems have been installed. This is similar to the maps that have been created to show where recharging systems have been installed. I would imagine that eventually any self driving car will need to know which roads have which level of infrastructure - a road that has only a single device communicating the wetness of the road, vs a road that has multiple devices detailing where other cars are, where pedestrians and bikes are, location of construction, can sense ice or water, or even can perhaps price lanes, or request cars to vacate areas so that emergency vehicles can get through will be a very important part of determining the route that future cars decide to take.
It would be good for the government to pull together metrolinx, transportation, planning, research, business and finance to start putting together some frameworks for these systems to be implemented in Ontario, and by Ontario.
Metrolinx and Government of Ontario need to start creating an environment where this can begin to be studied and developed in ontario. Leaving this to california, or germany is short sighted. There are a number of things that they should be doing in the short, medium and long term to encourage development of all three levels of automated driving which are -
1) warning and avoidance
2) controlled conditions (right of ways) - driver takes over when car is not sure
3) completely automated
Some of the things they should be doing:
1) Creating zones and corridors that are free to experimentation in the areas of car-to-car and car-to-infrastructure networking - for example, Waterloo in the area of the university - if people want to connect devices that communicate with vehicles to parking structures, street lights, light posts etc. there should be a very clear way of engaging the government to get that done. There are "fake" towns that are used by police services, driver training etc. that could be used in off hours for this testing as well. This will encourage research and development. It's also necessary to provide sections of highways, rural roads etc. Systems that solve the "warning and avoidance" item above could be intersections that communicate with cars to tell them if the light is red or green - and the car can then warn a user they may be about to drive through a red light. These systems should be fairly easy to implement and test.
2) All the transportation systems and boards should be required to provide open and free access to up to date data - on parking availability in Go Parking lots, on metered parking, etc. if the data doesn't exist or we can't get it, we should be challenging people to develop ways to get it and create it.
3) For number 2, we have a number of routes that are already protected (subways), and some routes that are partially protected (LRT/streetcars on spadina, lakeshore, st. clair, BRT's in vaughn etc.) there should be efforts made to make these routes as safe as possible using current technologies, but potentially to also upgrade them to fully automated where possible.
4) We also have a number of roads where lanes could be made partially protected or fully protected, or act in such a manner, I'm thinking areas of the 401 between windsor and waterloo, areas of the 400 north of parry sound->sudbury and beyond. I imagine that if the government offered a roadmap to self-driving freight on these routes that a number of companies would be willing to work on creating such a system. There could be a number of clear rules - trucks must be clearly labeled as self-driving, they must not operate in inclement weather, all systems must prove themselves with XXX,XXX of hours "attended" driving, they must pull-over if traffic becomes heavy, they must not be carrying hazardous material, they must not travel through cities automatically but park and be driven to their final destination. I would imagine that it would even pay for companies to outfit the road with censors to increase the speed they could deliver such a system (that check for incursions, or road/weather conditions). Work with fedex or a number of other companies that ship lots of stuff by truck.
5) We should be investigating systems that provide for less drivers of transit vehicles (bus trains etc.) that could be implemented without much difficulty in BRT conditions.
6) We should start to publish maps that show where infrastructure-to-car systems have been installed. This is similar to the maps that have been created to show where recharging systems have been installed. I would imagine that eventually any self driving car will need to know which roads have which level of infrastructure - a road that has only a single device communicating the wetness of the road, vs a road that has multiple devices detailing where other cars are, where pedestrians and bikes are, location of construction, can sense ice or water, or even can perhaps price lanes, or request cars to vacate areas so that emergency vehicles can get through will be a very important part of determining the route that future cars decide to take.
It would be good for the government to pull together metrolinx, transportation, planning, research, business and finance to start putting together some frameworks for these systems to be implemented in Ontario, and by Ontario.
Solid Snake
Active Member
Hopefully after what happened with the Chicago L Train accident at O'Hare Airport Station, the TTC will see the merit to have the train drive themselves just like in Montreal who's constantly one of the safest in the world
rbt
Senior Member
There are lots of ways of stopping a train mechanically and the TTC uses these.
Significantly more people than just the driver is at fault in Chicago, likely starting with decades of underfunding maintenance.
Last edited:
BMO
Senior Member
I've heard rumours, from sources inside the Provincial Government, that they are already building a draft plan for automated vehicles in Ontario so that they can be legally allowed on roads and planned for.
qwerty1
New Member
Time will tell if the automated car vision is Jetson like or not. I suppose I have more faith in the fully automated car vision if only because people frequently underestimate where technology will take us. Look back at some of the views of computer execs from the 50s, 60s and 70s about the role of computers in society. From Bill Gates saying something along the lines of people will never need anything more than 640k to the IBM dude who allegedly said there wouldn't be any need for more than 5 supercomputers in the U.S.
If they ever figure out how to reliably and stably store quantum information technology is going to go in inconceivable directions.
georgevicbell
Active Member
Sounds good...there are a lot of good jobs to be had in this area - and lots of money to be made....everything from computer science and engineering jobs to installation of devices on roads or retrofitting vehicles, rebuilding of roads for enhanced functionality...sales to many other countries, and a much more efficient transportation and freight system...
They should hold some conferences, and provide some funds for universities to start programs...along with development areas and targets...maybe even an x-prize type competition....first person to deliver a truck full of goods from Parry Sound to Sudbury with no driver intervention on a regular basis for a month, during the winter gets $10 million...
BMO
Senior Member
Best part about this, IMO, would be that most of the onus is on the private sector to bring forth the technology. When automated cars come out the advantages will be that they will require very little in the way of physical infrastructure to get up and going, and it behooves private automakers to implement the technology as quickly as possible to stem lagging car sales from congestion related transfers to other modes of transit and increased competition from startup manufacturers. I wouldn't be surprised if some aspect of automated cars are present within the end of this decade. It may simply start off as automated systems in parking garages that automatically drive and park your car for you, or a hybrid system of having the 401 series highways automated (less variables like lights and pedestrians, etc) and having the cars change back to manual once off the highway. How much cheaper would it be to reduce congestion if we didn't need to spend $30B on subway lines that take a long time to build? What if we could spend substantially less to retrofit our existing road infrastructure with sensors and a connected network that allows the city's light system to communicate with peoples' cars? Some of the biggest delays faced today on highways stem from human error or human tendencies. We all have our favourite routes, but are they the most efficient? What if we could trail the car ahead of us much closer freeing up a bunch of space for more vehicles? What if city light systems could efficiently disperse traffic throughout the entire system making the best use of our road capacity? What if there were fewer accidents and less construction(since cars traveling put less strain and weight on the asphalt)?
I'm really excited for the advent of automated driving, and I personally think it'll have one of the greatest impacts on future cities.
Last edited:
diminutive
Active Member
Well, people both under and over estimate technological change. We're still a long way off from 2001 A Space Odyssey, but someone who grew up during the era of Mercury, Gemini and Apollo may have thought that was a reasonable extrapolation. Technological change seems very non-linear.
I'd agree that automated driving technology has a promising future ahead of it. We should just be careful about making extrapolations. Maybe automated driving technology will mature in a way we're not even aware of...
Also, that Bill Gates quote is apocryphal. His actual quote was closer to: "My first microcomputer had 12K of memory. When I expanded to a full 64K, I thought I had all the memory I’d ever need. Hah. I know better now." It's almost tied with Marie Antoinette's "let them eat cake!" for most enduring misquotes in history.