Society of Robots - Robot Forum
Software => Software => Topic started by: Benn on February 03, 2010, 10:27:01 AM
-
Hy guys. I have a questions on Admin's Wavefront Algorithm, described in his tutorial.
I understood that this algorithm also make some kind of mapping. It is possible, after storing it in memory, to represent on the screen? So that i may see in which environment my robot was wandering?
Because i want to use this implementation in my robot, but first, i want to map the entire room, then sent it somehow(i'm not sure yet how exactly) to my laptop, and see it on the screen. And after this, i would tell him a goal where to reach.
Thanks.
-
fairly techie explanation here, try and follow it, but don't blow your brain up ;D
right, no in a conventional sense. unless you like matrix style outputs.(the movie or the math, take your pick) the robots map isn't really understandable at a basic level of interaction. add in low data transfer rates, and it becomes even more horrible.
the way around this is to pass data on the objects the robot detects to the pc. (boring techie detail)thus using low data transfer rates, and minimal power and processing capablities on the robot. the con of this, the computational cost to the pc is higher, and you duplicate some processes, increassing the overall computational cost of the system(end of boring techie detail)
in laymans terms, the robot can have a low powered processor, but the pc needs more processing power. so basicly, the pc needs to be running windows 98 or better :P pc's have masses of processing power. just add a few nano seconds to the processing time.
if you post process the data, you can get a vector representation of what it is that the robot can detect. you can then process that into a visual represntation of the data. the easiest way is to draw the vectors as lines on a 2d bit of "paper"
effectivly, you can't really get a robot to draw a map and send it to the computer. if it passed a vector representation of the objects it detects to the pc, the pc then does the image processing. the volume of data passed is a minimum. if it sees a giant wall in all of its sensor possitions, it passes a single vector for the whole wall. which is a lot faster than sending each sensor reading, or building a map and passing an image over wireless.
at some point you need to get into GUI's and image processing. I recomend C sharp, but whatever you want.
-
Have it output data using serial/USB to hyperterminal during each map update. You can easily get the map to look like below. It isn't as pretty as a SLAM map, but its functional and useful.
(http://www.societyofrobots.com/images/programming_wavefront_simulation2.gif)
-
hi.. it looks like your robot moving in the video provided is wireless when scan the path in the kitchen area...
so, how the robot store and transfer the data from sonar sensor after scanning its environment?..
sorry for the broken English..
-
hi,
have a look at here, it is my code, it will pass a 15x15 grid from the arduino to processing.
if you use the admins code like i did, there will be an array of the map, just pass that array.
http://arduino.cc/forum/index.php/topic,59074.0.html (http://arduino.cc/forum/index.php/topic,59074.0.html)
i have now got a nokia 6100 lcd working, to show the grid
here is my robot
http://letsmakerobots.com/node/26740 (http://letsmakerobots.com/node/26740)