Simulator Prototype

23 views
Skip to first unread message

Chris Hyde

unread,
Nov 14, 2013, 3:10:31 AM11/14/13
to palouse...@googlegroups.com
I showed off a little prototype to simulate the sub last Tuesday, and i will probably show it off at the Sunday meeting.  I figure I'd drop it off here if you guys want to run it and take a look at it.

I think this is worth pursuing, but I am not going to work more on it until next semester, as there is much other work to be done.  Happy to hear thoughts, concerns, or whatever. If we go ahead with this, lots of people will be getting their hands on it.

TO RUN:
- unzip
- open main01.html in chrome
main.zip
Prototype01_02.png

loganpevans

unread,
Nov 14, 2013, 4:36:42 PM11/14/13
to palouse...@googlegroups.com
This is quite awesome, and I think we should spend some time integrating this into the grapevine so that we have a quick turnaround environment for the vision -> AI -> movement command pipeline.

My confidence in the success of our robosub entry will shoot way up if we can integrate a simulator into our development environment.

Joel Wadagnlo

unread,
Nov 14, 2013, 6:56:21 PM11/14/13
to palouse...@googlegroups.com
I agree, being able to run our code in simulation will be vastly helpful.  It will help us find small bugs that we wont have to actually test, live on the sub.  Once it passes simulation tests then we can test it on the actual sub to find and correct any real-world problems.

loganpevans

unread,
Nov 14, 2013, 7:07:29 PM11/14/13
to palouse...@googlegroups.com
Since this is implemented in javascript, how difficult will it be to integrate this into the grapevine? Another way of putting that: how difficult is it to do ZMQ in javascript?

Chris Hyde

unread,
Nov 14, 2013, 7:39:39 PM11/14/13
to palouse...@googlegroups.com
Translating this to another language/API isn't out of the question.  This is more of a proof-of-concept.  That being said, pyOpenGL and most python 'API's' aren't very robust, and would require a LOT more work to get the same level.

loganpevans

unread,
Nov 14, 2013, 7:45:13 PM11/14/13
to palouse...@googlegroups.com
How hard would it be to export an image from the sub's view? Saving the image to a file should be sufficient, but you'll need to talk to Cameron to confirm that this will work.

Also, can you make the simulator respond to commands from a filestream (probably stdin)? If so, it won't be too hard to write a python bridge module that is aware of the grapevine, but can interface with your javascript program.

Chris Hyde

unread,
Nov 15, 2013, 4:50:11 AM11/15/13
to palouse...@googlegroups.com
from about 5 minutes of research, looks not *too* hard. I think we can get away with a server/client type socket. http://yz.mit.edu/wp/web-sockets-tutorial-with-simple-python-server/ not sure about video though..

Cameron Evans

unread,
Nov 25, 2013, 11:45:29 PM11/25/13
to palouse...@googlegroups.com
Whew, got to this thread a bit late. ZMQ has javascript bindings.
http://zeromq.org/bindings:javascript
Reply all
Reply to author
Forward
0 new messages