Re: [Torcs-devel] Bsc Project Ideas

[<wi...@pi...>]

On Tue, 07 Nov 2006 17:20:05 +0100, Lewis Foster <do...@gm...> wrote:

> Hi, my first post to the list.
>
> I'm looking at doing a physics type final year project for my Bsc  
> Computer
> Science. Torcs is looking nice and it would be great if I could put my  
> time
> into a racing sim to try and improve some aspect of the game.
>
> Does anyone have any ideas as to what I could try and improve?
>
> Also, what physics code does torcs use?
>
> Thanks,
>
> Lewis


Hi,

not really what you had in mind but you may like to consider this idea.

For some time I have been toying with the idea of making a braking  
distance awareness program based on Torcs.

Apart from being an intersting test to run at home it would be very useful  
in driving tution to make people aware of just how far behind you NEED to  
be to avoid a collision.

The simplest form is pretty much already doable: circulate at a normal  
road speed , then the instructor or a random event in torcs signals an  
emergency . You hit the brakes and then review your reactions in a play  
back.

Rolling start , empty road just to appreciate braking distances and  
reaction time to hit the brakes.

Program mods to take it further:

1/ on-screen display of speed, reaction time and breaking distance.
2/ child, dog , ball, pram, sudden car from dirt track... events.
3/ new urban tracks more suited to the context of driving. Addition of  
some urban robots (the torcs automatic cars) to make roads busier  
+ motorbikes, pushbikes.
4/ more urban car specs: clios and escorts rather than race tuned subarus.

To judge by most road driving, people do not realise they would not even  
have the time to close thier eyes and say "shit!" before hitting the car  
in front, let alone actually touching the brake pedal.

Although this may not be as effective as Tony Blair getting the police to  
falsify the road accident statistics , the lives saved would be real  
rather than statistical fiction.


In sticking to open source and with a couple of privisos, the end result  
could be distributed on a knoppix or similar live CD that could work out  
of the box without installation on most PCs.

There would be the need for a seriously scientific approach to measure car  
parameters that affect braking simulation and comparison with real car  
data. This will almost certainly lead to some applied physics and work on  
the modelling to provide meaningful results.


That's a fairly broad outline of what would be involved if that seems  
interesting or relevant to you.

regards.