Robothon RoboMagellan waypoints

5 views
Skip to first unread message

Paul Danset

unread,
Jun 3, 2012, 6:37:15 PM6/3/12
to Robo-Magellan Project
I've uploaded the Robothon waypoint data on Github here:


In particular, the waypoints overlayed over satellite imagery is here:


The official coordinates given to contestants are here:

To summarize, there was a start point and 4 pylons.  Two of the pylons had no line of sight.  The final pylon was way out of the way, around a building.   I've added two "virtual" waypoints on top of that because two of the pylons were only accessible from a certain direction (one was in a shrub area which confounded the other robot).  These two virtual waypoints are suggestions to the robot how best to approach the pylons in a safe way.

After all the discussion of how accurate the waypoints were, my guess was wrong.  It turns out that the waypoint data which we were given were just measured with an iPhone.  So while the precision of the data we were given -- 4 decimal places implying a resolution of ~10cm -- was only actually accurate to the normal ~3-5 meter range.  Accuracy vs precision.

So in hindsight, it's better not to rely on the handwritten coordinates and just use our own phones (preferably the same or similar phone used on the robot) to survey the pylon waypoints.  Our navigation app should just read those waypoints, plus the virtual waypoints.  The homing algorithm will have to take up the slack, correcting for the error in the GPS measurements.


Shubham Shukla

unread,
Jun 3, 2012, 7:22:47 PM6/3/12
to Robo-Magellan Project on behalf of pd
Hi Paul,

That's super helpful. I am planning on revisiting all the code I wrote in the past 6-8 weeks and cleaning it up. Let get together sometime soon and take this further.

-Shubham

-- You received this message because you are subscribed to the Google Groups Robo-Magellan Project group.
 
Check out our project page: http://www.jigsawrenaissance.org/donation/robo-magellan-project/
 
To post to this group, send email to robo-magel...@googlegroups.com.
To unsubscribe from this group, send email to robo-magellan-pr...@googlegroups.com.
For more options, visit this group at https://groups.google.com/d/forum/robo-magellan-project?hl=en

Reply all
Reply to author
Forward
0 new messages