Home
914 PC Bots Forum  


c6jones720
User

Platinum Boarder
Posts: 350
graphgraph
Karma: 3  
Survey results - 2007/05/24 22:57 This thread discusses the Content article: Survey results

Lets hope some of those goals can be accomplished for this community sooner rather than later
  | | The administrator has disabled public write access.
motters
User

Platinum Boarder
Posts: 132
graphgraph
Karma: 1  
Re:Survey results - 2007/05/28 16:57 Well I think I can do the vision/navigation part of the problem, but my main trouble is just one of acquiring mobile robot hardware upon which to run the software and do final integration testing.

I've got about as far as I can with simulated test runs using camera images taken from a tea trolley, and am reasonably confident that the software is going to work. Now it's just a hardware problem.

For details see http://code.google.com/p/sentience/

and for test data and animations see http://groups.google.com/group/sentience/files
  | | The administrator has disabled public write access.
jamesbruton
Admin

Admin
Posts: 683
graph
Karma: 8  
Re:Survey results - 2007/05/28 18:52 Hi Motters, It's good to know that you've got this far with your software.

Why not stick your laptop on the tea trolley and write a piece of software to display big arrows on the screen for left/right/back/forward so you know which way to push it when it asks you to?...

You might have a few issues with precise odometry, but I suppose it depends how good the vision guidance/SLAM is - maybe it can work out what's going on itself?

Otherwise, what are you options for a commercial unit that are up to the job? - short of building a mobile base of course... is a 914 basic unit out of the question?

http://store.pcpowerzone.com/pc9baun.html
  | | The administrator has disabled public write access.
c6jones720
User

Platinum Boarder
Posts: 350
graphgraph
Karma: 3  
Re:Survey results - 2007/05/28 18:56 Be assured there already is a scripting language and a speech recognition program for the robot too.
  | | The administrator has disabled public write access.
motters
User

Platinum Boarder
Posts: 132
graphgraph
Karma: 1  
Re:Survey results - 2007/05/28 19:27 Provided that I can get a decent odometry signal I'm pretty sure I can make it work. I'm not expecting that odometry will be perfect and that dead reconing error will accumulate in the usual way. The motion model can cope with this kind of uncertainty.

At the moment I'm doing something similar to what you describe, using a laptop on a tea trolley with some cameras. Manual guidance would be too inaccurate though - I need some actual measurement of change in pose/position which should be independent from the vision system for maximum reliability.

Another alternative, which would be the ideal, is to use inertial sensing. I have a 3 axis accelerometer which I've experimented with, but so far it looks like the output from this sensor is just too inaccurate compared to conventional odometry. The accelerometer is perfect for measuring things like tilt angle relative to gravity, but for measuring changes in position it seems that you need a very high sampling rate and some sophisticated noise filtering.
  | | The administrator has disabled public write access.