Page 3 of 6 FirstFirst 12345 ... LastLast
Results 31 to 45 of 80

Thread: Unlimited Class Race Car

  1. #31
    Join Date
    Feb 2008
    Location
    Wishing I was in Moscow, Idaho
    Posts
    2,585
    Quote Originally Posted by pimento View Post
    Putting an 'emtpy' car around an empty racetrack at speed has been done - there was an M3 on Top Gear that did that. I can't remember what the prep was for it though.
    A couple other people have done it as well, but never at F1 type speeds which is I think what they're arguing.
    Big cities suck

    "Not putting miles on your Ferrari is like not having sex with your girlfriend so she'll be more desirable to her next boyfriend." -Napolis

  2. #32
    Join Date
    Aug 2004
    Location
    IA
    Posts
    467
    Quote Originally Posted by wwgkd
    True. But the big chellenges in DARPA hasn't been the positioning software, it's been identifying and navigating around obstacles. The challenge isn't speed or precision, it's the large number of unknown obstacles scattered randomly around. Yes, they go slow, but that's because they don't have to go quickly, nor is a precise course an asset when you don't know the correct course ahead of time.
    You are exactly right, that is why it is different. The F1 version would be about speed and it would be about precision. So you can’t rely on the same solutions.

    Quote Originally Posted by wwgkd
    Identifying known objects at a high rate of speed has already been accomplished, even from unusual angles. And 3.520 Hz is not a hard target to match (even if it is just a lower bound.).
    Are you sure that is not a hard target to match. 3.5 GHz is just the calculation speed, not just processor speed. That means that all the data that needs to get to the processor has to be done so as fast if not faster to support the calculation above. Meaning the entire system must be able to support high speeds and low latency data movement.

    As well, are the technologies you are thinking about really well adapted to this application? I am sure they would work at 200 MPH but will they work at 200 MPH while 2 feet off the ground and 25 feet from the area of interest?

    Quote Originally Posted by wwgkd
    I still think the toughest part of this has nothing to do with going around a track quickly but lies in dealing with other vehicles
    Ignoring traffic for a moment driving an F1 car around the track is not about obstacle avoidance. Instead it is about having a good enough car model to be able to predict what the car will do next. I cannot imagine that will be a simple car model. I would suspect you would need all kinds of data such as engine speed, throttle position, tire speeds, pitch/roll/yaw data, air speed, brake position, tire temperature, tire pressure, fuel load, etc, etc, etc. All this data would have to get rolled into a car model that would predict where the car will be at the next instant and what corrections need to be made to get the car to where it should be. Keeping in my mind that many of the inputs are not linear and will be functions of time, temperature, speed, etc.

    I really don’t think it is a simple plug and chug type calculation. It would more likely require, and reserve the right to be wrong, a complex solution at every time iteration (I am imagining multiple partial differential equations).

    Quote Originally Posted by wwgkd
    I'm not so sure it is beyond us, or at least not far beyond. I'm pretty sure if you combined a number of existing technologies with a large budget you could get a single car to go around a course at a very high pace in a relatively short amount of development time. Again, this will be dealing with a known course and known obstacles.

    So you can map the course using any number of technologies with high precision. But once the care leaves the pit lane, how does it know where it is on the high definition map? GPS, only has accuracy in the ‘feet’. You could supplement with an inertial navigation system but that is subject to drift so it becomes less accurate the longer it is used, and total system error would then become proportional to the GPS error which was already stated as unacceptable. As already suggested you could use a visual imagining system but then you are talking about some real serious computational and graphical hardware. Not to mention you would probably need multiple high speed cameras to capture data fast enough (A 60 Hz progressive camera would only capture a new frame every 4.9 ft at 200 MPH). But then again maybe interpolation between 5 foot increments would be acceptable.

    Combining existing technologies is easier said then done, I mean that is how 90% of military hardware is sold and we all see how cost effective that is. I think it could be done with enough money, as I do of most things, but with today's technology I don't think it would be trivial.

    Another solution would be to line the track with unique markers/beacons that could be used to position the car, but I think that would be the least interesting and most slot car-ish solution.

    Quote Originally Posted by wwgkd
    The millitary puts a lot of money into things, but they also go with the lowest bidder that can fullfill the requirements.
    I would agree with everything but the part in bold, that usually just ends up being a ‘desirement’ ;-)
    "In theory, theory and practice are the same. In practice, they are not."

  3. #33
    Join Date
    Feb 2008
    Location
    Wishing I was in Moscow, Idaho
    Posts
    2,585
    Quote Originally Posted by Alastor View Post
    You are exactly right, that is why it is different. The F1 version would be about speed and it would be about precision. So you can’t rely on the same solutions.
    That was more a response to Matra saying that if they can’t be successful in the DARPA challenge then they probably couldn’t be here, either.

    Quote Originally Posted by Alastor View Post
    Are you sure that is not a hard target to match. 3.5 GHz is just the calculation speed, not just processor speed. That means that all the data that needs to get to the processor has to be done so as fast if not faster to support the calculation above. Meaning the entire system must be able to support high speeds and low latency data movement.

    As well, are the technologies you are thinking about really well adapted to this application? I am sure they would work at 200 MPH but will they work at 200 MPH while 2 feet off the ground and 25 feet from the area of interest?
    I believe that would be 3.5 KHz, rather than GHz.

    Obviously there wouldn’t be a straight transfer, but if you could process multiple sources of input at speeds well above 200 MPH and identify specific targets in real-time, I believe a solution could be found.

    Quote Originally Posted by Alastor View Post
    Ignoring traffic for a moment driving an F1 car around the track is not about obstacle avoidance. Instead it is about having a good enough car model to be able to predict what the car will do next. I cannot imagine that will be a simple car model. I would suspect you would need all kinds of data such as engine speed, throttle position, tire speeds, pitch/roll/yaw data, air speed, brake position, tire temperature, tire pressure, fuel load, etc, etc, etc. All this data would have to get rolled into a car model that would predict where the car will be at the next instant and what corrections need to be made to get the car to where it should be. Keeping in my mind that many of the inputs are not linear and will be functions of time, temperature, speed, etc.

    I really don’t think it is a simple plug and chug type calculation. It would more likely require, and reserve the right to be wrong, a complex solution at every time iteration (I am imagining multiple partial differential equations).
    I imagine many of those inputs (fuel load, etc) could be dropped, or at least they wouldn’t require real-time/constant updates. These sensors already exist for cars and aren’t particularly resource consuming.

    Quote Originally Posted by Alastor View Post
    So you can map the course using any number of technologies with high precision. But once the care leaves the pit lane, how does it know where it is on the high definition map? GPS, only has accuracy in the ‘feet’. You could supplement with an inertial navigation system but that is subject to drift so it becomes less accurate the longer it is used, and total system error would then become proportional to the GPS error which was already stated as unacceptable. As already suggested you could use a visual imagining system but then you are talking about some real serious computational and graphical hardware. Not to mention you would probably need multiple high speed cameras to capture data fast enough (A 60 Hz progressive camera would only capture a new frame every 4.9 ft at 200 MPH). But then again maybe interpolation between 5 foot increments would be acceptable.

    Combining existing technologies is easier said then done, I mean that is how 90% of military hardware is sold and we all see how cost effective that is. I think it could be done with enough money, as I do of most things, but with today's technology I don't think it would be trivial.

    Another solution would be to line the track with unique markers/beacons that could be used to position the car, but I think that would be the least interesting and most slot car-ish solution.
    If nothing else GPS can achieve an accuracy of about 2mm using carrier phase tracking, though I think you would probably also want to use visual imaging. Maybe sonar or radar? As mentioned markers or beacons could be used. Either way all the space currently taken up by a driver and the safety features for that diver have suddenly been freed up, so you have some room and weight to play with, and it would probably lead to some funky looking designs. Not saying success would happen instantly, but I don’t think it’s an impossible task to get it up and running soon. As was mentioned it has been done, albeit at lower speeds. You’d need some modeling for dealing with other vehicles in a series, of course, unless it was in a time attack format. And crashes would happen, same as in the racing we know and love today.

    The real problem is that no one would watch more than a couple times, and the money invested would never be worth it. Computers aren’t heroes, and even if you went the remote controlled route with full simulators and a driver to control the vehicle I doubt it would drum up much interest. It could be done, though.

    Quote Originally Posted by Alastor View Post
    I would agree with everything but the part in bold, that usually just ends up being a ‘desirement’ ;-)
    Well, them too.
    Big cities suck

    "Not putting miles on your Ferrari is like not having sex with your girlfriend so she'll be more desirable to her next boyfriend." -Napolis

  4. #34
    Join Date
    Oct 2006
    Posts
    10,227
    All I have to say is that I reject your use of the Imperial system in calculations.

    That is the system of Satan!

  5. #35
    Join Date
    Feb 2008
    Location
    Wishing I was in Moscow, Idaho
    Posts
    2,585
    Quote Originally Posted by Kitdy View Post
    All I have to say is that I reject your use of the Imperial system in calculations.

    That is the system of Satan!
    You're still butt-hurt over the betrayal of your ruler by Satan. Admit it!
    Big cities suck

    "Not putting miles on your Ferrari is like not having sex with your girlfriend so she'll be more desirable to her next boyfriend." -Napolis

  6. #36
    Join Date
    Oct 2006
    Posts
    10,227
    Quote Originally Posted by wwgkd View Post
    You're still butt-hurt over the betrayal of your ruler by Satan. Admit it!
    South Park, yeah?

    I... Just hate Imperial with a passion. I hate goddamn using it - when it comes to weight, I think in pounds, height (only human height) in feet and inches, and when it comes to cars, hp and lb-ft.

    Everything else is metric - the system of Jesus.

    Do you love Jesus or Satan?

  7. #37
    Join Date
    Feb 2008
    Location
    Wishing I was in Moscow, Idaho
    Posts
    2,585
    Quote Originally Posted by Kitdy View Post
    South Park, yeah?

    I... Just hate Imperial with a passion. I hate goddamn using it - when it comes to weight, I think in pounds, height (only human height) in feet and inches, and when it comes to cars, hp and lb-ft.

    Everything else is metric - the system of Jesus.

    Do you love Jesus or Satan?
    Hehe. Well I have to admit that using the Metric system would probably be easier if I had grown up with it.

    But since I didn't; it was good enough to conquer the world and forge an empire and it's good enough for me! It's called the Imperial system for a reason. The sun never set on the Brittish Empire, until they converted to Metric and I refuse to make the same mistake!
    Big cities suck

    "Not putting miles on your Ferrari is like not having sex with your girlfriend so she'll be more desirable to her next boyfriend." -Napolis

  8. #38
    Join Date
    Dec 2003
    Location
    nr Edinburgh, Whisky-soaked Scotland
    Posts
    27,775
    guys, can I remind what we were tyring to suggest was a car that coudl go FASTER than a driven one could manage.
    So making assumptions like not worrying about fuel load, corner weights, wind, downforce differences or distances of a few inches is just failing at the first fence.
    All of those things are being done by a driver to GET to the speeds that they struggle to cope with the physicality. So step one HAS to be a system whcih can actually take all of that into account in real time and come up with adaptive decision making on each lap at each corner. AND be accurate to less than an inch.
    "A woman without curves is like a road without bends, you might get to your destination quicker but the ride is boring as hell'

  9. #39
    Join Date
    Apr 2003
    Posts
    5,456
    I don't actually think having a computer controlled car driven at speed by itself will be that big of an issue. Most of the vehicle dynamic sim has a "idealized" driver in loop to do the sim. Provided they have detailed track data they can use that "driver" to drive the real car. The issue will be primarily detail work in tuning the multitude of actuators that puts the input of the computer driver in the actual vehicle. It'll require tuning, but it shouldn't be horrible. Running with traffic might be another deal altogether....The DARPA challenge is the fact that the computer is in effect learning the unknown as it goes, and the car has to deal with the stuff as it was presented to it. I think in this case for a closed course vehicle its more of an challenge of coping with the known. All that data you have mentioned are in effect taken into account at the level we are talking about in terms of doing a "sim". They run 500+ channels in F1 cars Data Acq for a reason...
    Last edited by RacingManiac; 09-27-2010 at 08:42 AM.
    University of Toronto Formula SAE Alumni 2003-2007
    Formula Student Championship 2003, 2005, 2006
    www.fsae.utoronto.ca

  10. #40
    Join Date
    Dec 2003
    Location
    nr Edinburgh, Whisky-soaked Scotland
    Posts
    27,775
    ^^ THAT is the point RM,
    The "idealized" part jst can't do real world.
    Hence why even the best race sim is a poor copy of the real task at hand.
    Let me repeat, the comparison with darpa was about the inability to actually map things particularly well even when given waling pace to do so. At the race level a track CHANGES with time, as more rubber gets laid it's grip improves. any dampness and those same bits "go off".

    Sims take a TINY fraction of the real world and a perfect example is NO SIM AS YET tales into account the slip ACROSS the width of a tyre and the effect it has on heat and wear adn grip. The math on tyre sim is well published.
    THEN you ahve the same on shock absorbers, springs, chassis twist etc etc et.
    Don't look at the MACRO cosm of the track but at the MICROcosm of the variations in each of the few inches of track at all times and that at no time is it exactly the same approach, travel over and exit on any.

    F1 cars run data acq on the INTERNAL factors and don't do ANYTHING with the vast majority of it in real time. Storage and PROCESSING are totally different tasks

    PROCESSING video to determine position is another horrendous task to try to do at speeds of 150+ mph.

    Sims are HUGE SIMPLIFICATIONS of real world and rely on gross averages.
    Don't use those as the benchmark.
    Instead go to the closed-loop-control requirements and realise how many variables are at playa nd how the driver "instinct" is at play on these. For example, is a front wheel understeering because it is passing an area of low grip, or has less weight on the tyre, or a bump, or going too fast or driver input has too much turn or different part of track, more lean in the chassis, etc etc etc I could list a thousand reasons why. Now the problem with control systems is they have to have rules to operate by - be the hard or fuzzy - and for each input you DOUBLE the decisions to make. So with 500 inputs then crudely you have 500 to the power 500 decision points ... SO then it's down to simplification BUT THEN the issue is which ones and when. Some will be ok some times and not others.

    It's not about "following" a 20 foot wide tarmac. It's about drivign it FASTER than a person coudl do remember
    "A woman without curves is like a road without bends, you might get to your destination quicker but the ride is boring as hell'

  11. #41
    Join Date
    Aug 2004
    Location
    IA
    Posts
    467
    Quote Originally Posted by wwgkd
    I believe that would be 3.5 KHz, rather than GHz.
    Doh….I am putting down my crack pipe now.


    Quote Originally Posted by wwgkd
    I imagine many of those inputs (fuel load, etc) could be dropped, or at least they wouldn’t require real-time/constant updates. These sensors already exist for cars and aren’t particularly resource consuming.
    But combining the data from all the sensor and doing ‘math’ with inputs is resource consuming.

    Quote Originally Posted by wwgkd
    If nothing else GPS can achieve an accuracy of about 2mm using carrier phase tracking, though I think you would probably also want to use visual imaging. Maybe sonar or radar?
    The carrier phase tracking is interesting although the 2mm accuracy is under ideal conditions, not always guaranteed, but still better than tradition GPS methods. Also, it doesn’t look like there is a lot of commercial hardware available and it looks like it requires some robust software to work at it’s best.

    I also like the idea of radar and sonar there might be some interesting opportunities. I don’t know how well sonar would work in air, but both technologies are similar. It seems like to you could use them to generate an ‘image’ of the track but also modify that image to create a kind of velocity map using the Doppler effect. However, I would expect both of these technologies to be more expensive than visual imagining because they both need a transmitter and receiver, whereas with visual imaging just needs a 'receiver'. Nevertheless, visual imagining and radar, not sure about sonar but it is probably the same, are horrendously computationally expensive.


    Quote Originally Posted by wwgkd
    The real problem is that no one would watch more than a couple times, and the money invested would never be worth it. Computers aren’t heroes, and even if you went the remote controlled route with full simulators and a driver to control the vehicle I doubt it would drum up much interest. It could be done, though.
    Agreed it would be cool once, maybe as a PR stunt, but I wouldn’t pay to watch it.
    "In theory, theory and practice are the same. In practice, they are not."

  12. #42
    Join Date
    Dec 2007
    Location
    └A & Connecticlump
    Posts
    5,367
    In terms of raw processing power or memory, surely that could be solved by outsourcing the computing to one of higher capacity in the teams' party buses and filling the car with sensors. The main problem I see with that is that the car would need a fairly instantaneous link to said party bus that is capable of sending an extremely large amount of data, which I think is where most of the (non-algorithmic) difficulty would lie.
    The computer could receive this glut of data, perform hocus pocus, and spit out elemental commands to the car; steer, brake, lean fuel mix, "Man the harpoons!" etc...

    Quote Originally Posted by Matra et Alpine View Post
    ^^ THAT is the point RM,
    The "idealized" part jst can't do real world.
    Hence why even the best race sim is a poor copy of the real task at hand.
    Let me repeat, the comparison with darpa was about the inability to actually map things particularly well even when given waling pace to do so. At the race level a track CHANGES with time, as more rubber gets laid it's grip improves. any dampness and those same bits "go off".

    Sims take a TINY fraction of the real world and a perfect example is NO SIM AS YET tales into account the slip ACROSS the width of a tyre and the effect it has on heat and wear adn grip. The math on tyre sim is well published.
    THEN you ahve the same on shock absorbers, springs, chassis twist etc etc et.
    Don't look at the MACRO cosm of the track but at the MICROcosm of the variations in each of the few inches of track at all times and that at no time is it exactly the same approach, travel over and exit on any.
    While the technology may not currently be available it seems like a large amount and diversity of sensors would be able, given F1 budgets, to see the track a few feet ahead and remember how it has changed in the last few laps. After all, because the real world isn't a simulation, it would merely be relying on a computer to process external input not to actually create a scenario. Like I said about sending data to the car and back, this may not be immediately available, but given F1 budgets and a few years or the normal pace of technology and a decade or so, it doesn't seem too unfeasible.

  13. #43
    Join Date
    Apr 2003
    Posts
    5,456
    Matra, you are vastly underestimating I think the capability of the computer system. Even the ECUs and whatnot at F1 level(on the vehicle) are still not that powerful, because they don't need to be. As you've said, mostly they are taken for sampling and storage. However if now you need to build an ideal driver to process those input, you'll be needing a lot more processing power. But even at that point, we have a vast amount of processing power thats far beyond most automotive level stuff in regular computing. An average human reaction time is a quarter of a sec, that 4Hz, Your desktop now has probably 4 processor with a capacity of 2+ GHz each. And its not all the events are reactive based, much like a driver learning the track, once he got the basics figure out, there are a lot of the stuff becomes "training" based, that he does not need to think about. That is the basis of the initial "programming" of the control. And once that is done the computer based driver should have capacity to deal with reactive based events. Much like the driver drives with the seat in the pants feeling on what the car is doing, the computer should also be able to identify events, based on input and sensor data to "know" what is happening. The correction should be no different to any other control system based control. Some of this is already in cars anyway to "assist" the driver. Think something like Mitsubishi's AYC, the driver has his input in his throttle, brake and steering, the AYC controls torque distribution via electronic diff with input on accelerometers, yaw rate sensor, steering position and throttle position, plus it can brake individual wheels. That is the vehicle sensing what is happening(or rather, its best guess), and applying controls to the situation. While there are no set "lines" for the car to follow per se, based on the setting the vehicle has a set threshold that it tries to maximize.

    The technology is there, and applying to something like a "circuit racing" is much more work in tuning existing methods rather than trying to do something thats breaking new ground. While no one is necessarily doing this, does not mean I think that it can't be done. A robotic driver that drives on public road that can find its way from a to b might be more work I think than making a car that can drive around the track as fast as it can....Computer is not good at making predictive action or judgement, but it can over come that on a circuit based on sheer amount of calculation and processing power, provided that your actuator can operate at a speed that is not much slower than it is at.

    Much much more difficult, a robotic WRC car....
    Last edited by RacingManiac; 09-27-2010 at 01:16 PM.
    University of Toronto Formula SAE Alumni 2003-2007
    Formula Student Championship 2003, 2005, 2006
    www.fsae.utoronto.ca

  14. #44
    Join Date
    Aug 2004
    Location
    IA
    Posts
    467
    Quote Originally Posted by f6fhellcat13
    In terms of raw processing power or memory, surely that could be solved by outsourcing the computing to one of higher capacity in the teams' party buses and filling the car with sensors. The main problem I see with that is that the car would need a fairly instantaneous link to said party bus that is capable of sending an extremely large amount of data, which I think is where most of the (non-algorithmic) difficulty would lie.
    The computer could receive this glut of data, perform hocus pocus, and spit out elemental commands to the car; steer, brake, lean fuel mix, "Man the harpoons!" etc...
    This biggest problem with that would be maintaining the high speed data link, what happens if the link is lost or degraded to the point where not enough data can be transferred? The next issues are going to be bandwidth and latency, both of which are limited compared to wired transmission. One way to get around the bandwidth limits is to compress the data, but that typically increases latency. Still an interesting idea.


    Quote Originally Posted by RacingManiac
    The correction should be no different to any other control system based control. Some of this is already in cars anyway to "assist" the driver. Think something like Mitsubishi's AYC, the driver has his input in his throttle, brake and steering, the AYC controls torque distribution via electronic diff with input on accelerometers, yaw rate sensor, steering position and throttle position, plus it can brake individual wheels. That is the vehicle sensing what is happening(or rather, its best guess), and applying controls to the situation.
    But in that case isn’t the car just taking the X that the driver chose and plugging it into an equation and finding the correct Y. That is relatively easy even with very complex equations. The problem is when you have some Y (output) and want to know what the correct X (input) is to get generate that output. Then you have to find the inverse function and that can be very tricky. Without the driver the car only knows what the next correct position of the vehicle should be, it must then calculate backwards to find the right input. Or does the car just guess and check itself down the track? I ask because I never studied ‘controls’.
    "In theory, theory and practice are the same. In practice, they are not."

  15. #45
    Join Date
    Feb 2008
    Location
    Wishing I was in Moscow, Idaho
    Posts
    2,585
    Quote Originally Posted by Matra et Alpine View Post
    guys, can I remind what we were tyring to suggest was a car that coudl go FASTER than a driven one could manage.
    So making assumptions like not worrying about fuel load, corner weights, wind, downforce differences or distances of a few inches is just failing at the first fence.
    All of those things are being done by a driver to GET to the speeds that they struggle to cope with the physicality. So step one HAS to be a system whcih can actually take all of that into account in real time and come up with adaptive decision making on each lap at each corner. AND be accurate to less than an inch.
    Drivers don't check fuel weight at each corner, thus something like that could be done at a much slower pace since the small useage between corners isn't going to be significant compared to the vehicle weight. Active downforce would obviously be a bonus (so many racing teams are trying to find some way of getting around that rule that it kind of has to be) and coordinated in with other systems. Much of this has already been done with some of the more advanced traction control systems, you wouldn't be doing ground breaking work in many of the areas related to maximising traction and "knowing" what each wheel is doing.

    Also, the distances would be 2mm, which is less than .08 inches. Far more accurate than what a driver accomplishes.

    Quote Originally Posted by Alastor View Post
    Doh….I am putting down my crack pipe now.


    But combining the data from all the sensor and doing ‘math’ with inputs is resource consuming.


    The carrier phase tracking is interesting although the 2mm accuracy is under ideal conditions, not always guaranteed, but still better than tradition GPS methods. Also, it doesn’t look like there is a lot of commercial hardware available and it looks like it requires some robust software to work at it’s best.

    I also like the idea of radar and sonar there might be some interesting opportunities. I don’t know how well sonar would work in air, but both technologies are similar. It seems like to you could use them to generate an ‘image’ of the track but also modify that image to create a kind of velocity map using the Doppler effect. However, I would expect both of these technologies to be more expensive than visual imagining because they both need a transmitter and receiver, whereas with visual imaging just needs a 'receiver'. Nevertheless, visual imagining and radar, not sure about sonar but it is probably the same, are horrendously computationally expensive.


    Agreed it would be cool once, maybe as a PR stunt, but I wouldn’t pay to watch it.
    I have to admit I stole the sonar idea from Ford, who uses it on their parallel parking system. They say highfrequency sonar works better than the radar systems that have been used before. BMW came out with a similar system immediately following Ford so there's probably something to that.

    Even as an ideal, 2mm is going to be better than what a driver can do, so a slight degredation would be acceptable, I think. But it should probably be worked on by engineers wearing black socks with sandals for future improvement.

    Also, at some point you wouldn't be relying on going around a track on a preprogrammed course, but just using that for an approximate placement and letting the cars other systems maximize your speed.

    Quote Originally Posted by RacingManiac View Post
    Much much more difficult, a robotic WRC car....
    This. Too much slippage and you get back into the "unknown obstacle" thing as the course changes much more drastically on dirt than asphault.

    Quote Originally Posted by Alastor View Post
    This biggest problem with that would be maintaining the high speed data link, what happens if the link is lost or degraded to the point where not enough data can be transferred? The next issues are going to be bandwidth and latency, both of which are limited compared to wired transmission. One way to get around the bandwidth limits is to compress the data, but that typically increases latency. Still an interesting idea.
    With F1 resources why not use their solution of multiple channels? Although controlling the cars through something so easily effected by radioshack gadgets may not work...

    Quote Originally Posted by Alastor View Post
    But in that case isn’t the car just taking the X that the driver chose and plugging it into an equation and finding the correct Y. That is relatively easy even with very complex equations. The problem is when you have some Y (output) and want to know what the correct X (input) is to get generate that output. Then you have to find the inverse function and that can be very tricky. Without the driver the car only knows what the next correct position of the vehicle should be, it must then calculate backwards to find the right input. Or does the car just guess and check itself down the track? I ask because I never studied ‘controls’.
    Well I think what you could do would be to used the "navigational" side of things to figure out where you want the car to be ideally, then use something like a more advanced version of traction controll to accomplish that. Ferrari's 599 XX has a pretty advanced system already, and navigational systems have been made to work already, why not just advance them both even further and then blend them together?

    Edit: Obviously I realize that's a huge simplification of the process, so don't go bothering me with reality.
    Last edited by wwgkd; 09-27-2010 at 10:24 PM.
    Big cities suck

    "Not putting miles on your Ferrari is like not having sex with your girlfriend so she'll be more desirable to her next boyfriend." -Napolis

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Similar Threads

  1. Favorite James Bond Automobile
    By toyota_trevor in forum Car comparison
    Replies: 110
    Last Post: 05-26-2021, 07:18 AM
  2. Famous Touring Cars
    By motorsportnerd in forum Racing forums
    Replies: 353
    Last Post: 07-04-2010, 08:57 AM
  3. Lotus Evora Cup
    By Ecnelis in forum Matt's Hi-Res Hide-Out
    Replies: 4
    Last Post: 02-11-2010, 11:38 AM
  4. 2007 Bathurst 12 hour
    By motorsportnerd in forum Racing forums
    Replies: 69
    Last Post: 05-13-2007, 12:24 AM
  5. V8 Supercars: Round 6 Hidden Valley, NT
    By fpv_gtho in forum Racing forums
    Replies: 2
    Last Post: 07-01-2005, 03:08 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •