Real raw images...?

8 posts / 0 new
Last post
casramsey
Real raw images...?

Hi Guys - is it possible for formula pi to supply us with a stack of raw images from a monster borg doing a few laps around the actual track? From today's test it was clear that the simulation images are not close enough to actual raw images for my team to do what we want, so it would be great if you could send a monster borg around the track, just on the standard code (maybe a couple of lane changes to mix it up a bit), and drop every image into a zip file we can get hold of? Or maybe you already have this somewhere?

Cheers

Arron Churchill
Arron Churchill's picture
Raw images

We do want to do this, so far we have not found the time :)

We will try and squeeze some time in to do a run recording stills, but it will probably be sometime next week.

casramsey
That....would be awesome!

That....would be awesome!

Thomas
Just my opinion on this...

I really thought twice posting this:

Providing raw images would help a lot for sure.
But it takes away one aspect of the whole competition: gathering yourself all valuable information to improve your race. Another point is that the competition is already running and I have for example already spend much time in this step, though I failed up to now (really unexpected problems, I can share afterwards).

just my two cents

casramsey
Fair point, but...

@Thomas - appreciate what you're saying, but I think if we're programming for driverless vehicles we should be using all the "real" information we possibly can. Like you, I have spent much time on this step, but find when we go on track for testing, the visual information is producing very different behaviour to the visual information in the simulation. Be good if we could all get the same on-track images to be able to tweak algorithms with real data, rather than guess at it.

Admittedly I probably should have collected images during the first testing round, but given there is no time between testing 0 and code submission for testing 1, I probably couldn't have used it to any concrete effect anyway. Hence I think a standard deck of real images from 4 or 5 laps round the track, for monster and yeti, as part of the downloadable race code would be really helpful to all.

(but also just my two cents ;o)

Jamie
Jamie's picture
Raw images

Just thought I'd weigh in on this. Yes, it would be nice to have a standard set of images to work from. But this is why I (and I'm sure I'm not the only one) spent a lap in Testing 0 gathering that data. Which may or may not have been at the expense of lap times. Not wanting to give too much away :) Giving out a set of images now, after the season has started does seem a little unfair to those who gathered that data themselves.

I agree that there wasn't much time to test with the new data, but there are plenty of testing rounds still to come. You could even take a gamble and try gathering data during a race and have it ready for the next testing session.

Ultimately, it's up to the folks at Formula Pi to decide whether or not to release images now or wait until the winter season. There are pro's and con's to each.

That's just my two cents. :)

Timothy Freeburn
Timothy Freeburn's picture
Library of track images from real robot

Hi All,
We always intended to put up images, and we will do shortly. The problem is up until today, we still had prototype MonsterBorgs running around the track. Recording images from the prototypes wouldn't have been accurate and could have led to confusion, but in hindsight they probably would have been useful.

I appreciate some of you have taken snaps and logs. Whilst useful to get started, I don't think it will be a huge advantage having them, especially as the prototypes were not calibrated.

The idea behind running the Testing round 0 - was to make sure people could handle the FTP transfer, get aquainted with how the process works, and to help us try and get our system sorted out (we still made loads of mistakes!) without any of the results being recorded.

There wasn't a huge amount of time between testing 0 and testing 1 submission, so I don't think much advantage should have been gained. In any case, we will put up a library of images from the real calibrated Monsters shortly, which everyone can use.

We hope to do a small blog post explaining the images, which hopefully will aid in everyones understanding.

Please keep feedback like this coming in, it is really useful for us to understand. There are loads of things we would do if we had more time, unfortunately finding time seems to be like looking in a forest for a Yeti*.

*The mythical creature version, not the robot!

Arron Churchill
Arron Churchill's picture
Actual footage

We have some actual footage and images up now, plus a bit of analysis of the real images verses the simulation: MonsterBorg self-driving footage.

It is only about two and a half minutes long (plus a long wait for me at the start), but it does cover most of the real world problems.

Add new comment

Filtered HTML

  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd> <pre>
  • Syntax highlight code surrounded by the <pre class="brush: lang">...</pre> tags, where lang is one of the following language brushes: bash, cpp, perl, python.
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Comment Images
Files must be less than 10 MB.
Allowed file types: png gif jpg jpeg.
Comment Attachments
Files must be less than 10 MB.
Allowed file types: txt pdf nfo doc docx rtf jpg png gif bmp zip tar gz csv xls.
CAPTCHA
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.