robotics (anytime D *)

[RoboCop01]

I am playing with the "Anytime D Star on an Occupancy Grid.lvproj" in "labview\examples\robotics\Path Planning\ADstar\ADstar on Occupancy Grid" directory.

I want to make it so that the robot size is more than one cell (4 cells to be exact). Is this possible?

Cheers,

Jono

[S_Kamath]

No, the AD* implementation assumes the robot occupies 1 cell. Can you tell me a bit about why a multi cell representation of the robot would be useful?

[RoboCop01]

We are competing in the NI Autonomous Robotics Competition run by NI Australia.

Here is the page for it: https://decibel.ni.com/content/groups/ni-arc.

The competion involves designing an autonomous robot to navigate a course (example course attached) by avoiding obstacles to pick up cubes and drop them off at certain locations.

The course has a 250 mm x 250 mm grid. Obastacles take up 2 of these grid cells.

Our robot takes up more than one cell (it is about 300 mm x 350 mm) so if we make the robot take up only one cell in the model we will end up driving into obstacles.

I have thought about increasing the size of each cell but then the obstacles would take up less than one cell each and we would be restricting oursleves from parts of the course that are free of obstacles.

Any help would be appreciated.

Cheers,

Jono

[S_Kamath]

Interesting competition, run by NI too!

The only method I can think of to adapt the grid is to block gaps that are too small for your robot via your mapping code. When trying to go through a gap if your sensors determine it is not large enough for the robot then mark it an obstacle and replan. A gap that is not large enough to safely navigate through is not a gap at all.

More work but perhaps more rewarding is to create a directed graph instead. You would be looking at creating grid points you can drive to "easily" from your current location, then a local obstacle avoidance algorithm can take on any small obstacles on the way. Most robots working in larger environments cannot afford to path plan on finely grained maps and take similar global\local approaches.

[Dwayne_MSD]

Hi... I'm new with LabVIEW and interested with this AD* sample program. Anyone, please help... I would like to apply this program to real robot with ultrasonic sensor. Which part should I change to gain signal from sensor and which part should I connect to motors. Very much thanks.