Time-Warp Machine control!

[docel]

Hello!
Immersed in a chat with the local kids, I surfaced with a weird notion: How to control a vehicle that responds to commands 5mins later? or 1 hour later??

Now this is what NASA and other intelligent beings there with haloes do all the time with the Mars Rover ( with a lot of help from a few super-Kids, I hear ). But, imagine you and me and lesser kids doing this in real life?

What fun to control this 'lazy' car around a particular circuit!!

 

[Nigel Goodwin]

Hello!
Immersed in a chat with the local kids, I surfaced with a weird notion: How to control a vehicle that responds to commands 5mins later? or 1 hour later??

Now this is what NASA and other intelligent beings there with haloes do all the time with the Mars Rover ( with a lot of help from a few super-Kids, I hear ). But, imagine you and me and lesser kids doing this in real life?

What fun to control this 'lazy' car around a particular circuit!!

You don't "control" it, it's self controlled with a high degree of autonomy with complicated sensors. Instructions are only sent in a fairly vague form, which it then interprets itself - if it meets something it can't cope with it will then stop and wait for instructions.

For an example, of a Mars rover, send it an instruction to proceed to an area 11.5m south, and take a soil sample - the robot will then decide itself how to carry out the instruction, even is it has to move around an insurmountable obstacle to do so.

 

[docel]

Right Nigel!
Thats for the Rover.
I was thinking about a small car with a time-delay, then everything goes a little cranky. If i want to make the car respond to my control, I have to wait for some time, say 5 mins. Every subsequent command will take 5 mins for a response. This is very confusing to the brain.
Given a course to navigate, one has to think forward. We are not used to this kind of navigation! It should be fun, and some experience too, to make one as an experiment, don't you think?

 

[akg]

so r u thinking of a RC Car that reponds to commands given 5 min before ?

 

[docel]

Right! I'm also thinking of a possible competition. The learning curve is different for age levels. Take for example the Remote control; the kids take to it like a fish to water whereas the adults fumble. Similarly control buttons etc.,
How would pre-learning/pre-training compete with impromptu learning?

I look forward to your views.

If the topic is out of context, the Admin/ Mod. may please shift this elsewhere appropriate?

 

[Nigel Goodwin]

Right! I'm also thinking of a possible competition. The learning curve is different for age levels. Take for example the Remote control; the kids take to it like a fish to water whereas the adults fumble. Similarly control buttons etc.,
How would pre-learning/pre-training compete with impromptu learning?

I look forward to your views.

It's a non-starter - you can't drive a real car, never mind a model one, in that way - you need feedback just in order to keep it going straight, and the delay time means feedback isn't possible.

It's easy enough to test, do you have a remote controlled car? - set out a straight course, with a bend after a number of metres - then try and drive the car with your eyes closed, and see what happens. Note that this will be CONSIDERABLY easier than a 5 minute delay, as it's only a couple of seconds!

 

[docel]

Yes, Nigel, learning has a lot to do with feedback. The brain responds to feedback from the senses. Time delay will create havoc by the time action is taken. This is overcome by reflex action via the synapse. The hand doesnt wait till the brain understands the nature of stimulus. It responds first!

Feedback is essential for the brain. The brain is conditioned by learning. Please understand my problem: if you put 2 people to compete on a similar track, at the same time, how would they fare? Mind you, this is just a technical game! On the human side, difficult to predict. But with practice, it may be easy?

And,
Is it possible to design a system that can work with a time delayed feedback mechanism?

 

[Nigel Goodwin]

Yes, Nigel, learning has a lot to do with feedback. The brain responds to feedback from the senses. Time delay will create havoc by the time action is taken. This is overcome by reflex action via the synapse. The hand doesnt wait till the brain understands the nature of stimulus. It responds first!

Feedback is essential for the brain. The brain is conditioned by learning. Please understand my problem: if you put 2 people to compete on a similar track, at the same time, how would they fare? Mind you, this is just a technical game! On the human side, difficult to predict. But with practice, it may be easy?

It doesn't matter about the people, it's all about the car - it has to be autonomous to be able to survive without instructions. If the only feedback mechanism is human, via the remote control, and delayed by five minutes, then it wouldn't survive to even receive the first instruction.

EXAMPLE: the car starts drifting to the left, you apply correction, but it takes 5 minutes for the instruction to be received - how far left has the car gone?, how many circles has it turned?, how many buses have squashed it?, how many holes has it fallen down.

And,
Is it possible to design a system that can work with a time delayed feedback mechanism?

Yes, I would say so, but NOT something like this which needs continuous and fast response to even maintain the status quo (such as moving in a straight line).

Like I said before, get a radio control car and try it blindfolded - just the same, but easier as the delay will be far shorter (so much less time for things to go wrong).

 

[Dr.EM]

I would assume the easiest way to implement it is by inserting delays after the switches in the remote control. Make sure they are switches though, not linear controls as that would be far more difficult to delay.

 

[DirtyLude]

I'm not really getting this either.

Nigel seems to think this car is going to be moving forward until receiving a command, but I think the original poster is thinking forward will be a command. Therefor a simple delay wouldn't be sufficient for the remote, since it would need to buffer all commands and send them in order with the correct offset delay that they were sent in.

Therefor the experiment would go something like: user will send command forward for 2 seconds, left/forward for 1 seconds, forward for .5 seconds, right/forward half speed for 1.5 seconds.... So that way 5 minutes after the first command is sent, the car starts moving forward. 5 minutes and 2 seconds after the first command is sent, the second command kicks in, 5minutes and 3 seconds and the third command, and so on.

In this way it's simply a matter of trying to figure out what buttons or commands to send in what order at what timings to get the car to go around the track. Unless this is a really slow track, 5 minutes seems like a long time. Seems easier to just log all the commands somewhere then replay them when you want the car to move.

 

[Nigel Goodwin]

I'm not really getting this either.

Nigel seems to think this car is going to be moving forward until receiving a command, but I think the original poster is thinking forward will be a command. Therefor a simple delay wouldn't be sufficient for the remote, since it would need to buffer all commands and send them in order with the correct offset delay that they were sent in.

Yes, if it isn't moving already, then all you're doing is asking it to complete a pre-programmed series of instructions. The question was about 'controlling' it, not executing a series of preloaded commands.

Therefor the experiment would go something like: user will send command forward for 2 seconds, left/forward for 1 seconds, forward for .5 seconds, right/forward half speed for 1.5 seconds.... So that way 5 minutes after the first command is sent, the car starts moving forward. 5 minutes and 2 seconds after the first command is sent, the second command kicks in, 5minutes and 3 seconds and the third command, and so on.

In this way it's simply a matter of trying to figure out what buttons or commands to send in what order at what timings to get the car to go around the track. Unless this is a really slow track, 5 minutes seems like a long time. Seems easier to just log all the commands somewhere then replay them when you want the car to move.

But, as I've I've said all along, you still need feedback, either internally in the vehicle, or via the remote control - a mechanical vehicle won't go straight unless there's something to keep it straight - unless everything is 100.000000000% perfect, including the track it runs on!. The slightest difference in loading either side will throw it off track, but more likely slight differences in the vehicle will throw it off - people can't even walk in a straight line with their eyes closed!.

 

[DirtyLude]

You're right, 5 minutes turns this more into a series of preloaded commands than control, unless the user wants to sit and wait 5 minutes to see what their commands have accomplished before issuing a new command.

You are thinking that a track has to be some kind of constrained course, when it could just as easily be a series of way points. Exact straight lines are not required, just hitting a particular mark or location is needed to continue.

I think as a perception experiment 5 minutes is way too long. I would think seconds delay would be adequate enough to throw off the person controlling and still allow things to keep moving. Given enough time and practice the user would be able to anticipate when to send the commands based on where the car will be after the delay is over.

 

[Roff]

There's another way of looking at this, where feedback still works.
For the sake of argument, let's say that my reaction time in the real world is 0.25 seconds. I decide to tell my car to stop, and .25 sec later, I apply the brakes.
OK, in this game we have a situation where my reaction time is 1200 times as long (5 minutes). If, instead of driving my RC car at 12 meters/sec max, I now drive at 1 cm/sec max, and everything else in my new universe is similarly slowed down, I can still cope with it (with practice). My RC car only goes 3 meters before I can react. Then the game becomes like NASCAR racing. How fast can you go before you crash? 2 cm/sec? 3?
Unfortunately, my life will only seem to be about a month long if we play the game for my entire lifetime.

 

[Oznog]

Doing a time warp is easy.
It's just a jump to the left.
and a step to the right...

 

[Papabravo]

Go Meatloaf!

Let's do the time warp again.....

 

[DirtyLude]

Now I have that song in my head.

 

[docel]

Hello folks! Thank you for taking the time and all the excellent views.

Yes, if it isn't moving already, then all you're doing is asking it to complete a pre-programmed series of instructions. The question was about 'controlling' it, not executing a series of preloaded commands.

The vehicle should move only by commands.

But, as I've I've said all along, you still need feedback, either internally in the vehicle, or via the remote control - a mechanical vehicle won't go straight unless there's something to keep it straight -

A moving wheel or a bycycle will continue to move, unless it hits an obstacle. Depending on the kind, it may continue to move , maybe in a different direction. A car will take the resultant vector of all the possible obstructions. Will this happen due to Feedback or Force?

You are thinking that a track has to be some kind of constrained course, when it could just as easily be a series of way points. Exact straight lines are not required, just hitting a particular mark or location is needed to continue.

This was something I had'nt thought of, Thanks Mark.

The slightest difference in loading either side will throw it off track, but more likely slight differences in the vehicle will throw it off - people can't even walk in a straight line with their eyes closed!.

Suppose the steering is spring loaded and centered?

We tried this, again. Immediately reading your post, I woke up some of my students and asked them to walk blind-folded. Then I asked them to "memorise" the terrain and try. They reported better success with the memory excercise.
( I did not do this myself as I'm good at this by training )

This is what I would to to test: Behaviour, Memory and Anticipation. Can they be ' intutive' or conditioned' ? and what possible method would suit the Human nature.

You're right, 5 minutes turns this more into a series of preloaded commands than control, unless the user wants to sit and wait 5 minutes to see what their commands have accomplished before issuing a new command.

Correct!

I think as a perception experiment 5 minutes is way too long. I would think seconds delay would be adequate enough to throw off the person controlling and still allow things to keep moving. Given enough time and practice the user would be able to anticipate when to send the commands based on where the car will be after the delay is over.

Maybe. Action-Perception cycle is fundamental to all intelligent creatures. When the agent acts, it interacts with the environment and 'changes' things or how it 'perceives' it. Behaviour will tend to modify this Action-Perception cycle according to 1) a predetermined sequence 2) an unpredictable action or 3) an action determined by its 'state of mind'
Eg: A squirrel may either run or oscillate between hunger and fear when a threat stands between food and itself.

With a 5 minute delay, How difficult will it be ? and what could be the resulting actions?

This is what I wish to determine. I wish to build and fit this system to one of my Robots as an experiment, if possible. This could be a collaborative group project, Research If you may!