Controlling the travel threshhold in A*

[craig_dickson]

We're building a model using A* controlling task executers to deliver flow items to queues.

How can I control the travel threshholds on the queues? I need the items to be dropped off actually at the queue, but the threshhold is much wider than that that, covering a space much larger than the queue itself. I really only want one node, or at most one grid space around it, as opposed to at least three (see screenshot.)

[joseph_g3]

@Craig DIckson

Queues shouldn't have travel thresholds that big. Could you share your model so that we can get a better look at the problem?

[craig_dickson]

@Joseph Gillespie

a-star-agv-test.fsm

[craig_dickson]

FYI, this is just a 'scratch' model to test how A* works and see if it is suitable for real model.

[joseph_g3 · Accepted solution]

@Craig DIckson

I don't know why I didn't think of this earlier, but your problem is actually pretty simple. You said that you want the task executors to drop off items at the queues, but that is impossible as long as you have your queues inside of a barrier. Barriers are designed so that task executors cannot enter them, so you will either have to edit or change the shape of the barrier to allow the task executors to reach the queue or you will have to accept them depositing items from a distance. As long as your queues are inside of an impassible barrier, the task executors will not be able to reach them.

[craig_dickson]

@Joseph Gillespie I don't think it's quite that simple. Even if I put the queue in open space, the agvs will deliver to several points on the queue's periphery, not to the single location of the queue, which is what I want. What that means is that they often try to deliver in the main travel path, not in the little safe harbor. I moved the queue into the barrier so that none of its points were in the main path, but sometimes it still made a lot more points as you saw.

There is not a way to control the size of the travel threshold so that things are delivered to one and only one point?

[philboboADSK]

If you want to travel to a specific point, use a Travel to Loc task to travel to the queue's location rather than a Travel task to travel to the queue object.

If you travel to a specific location, then it will go to that specific location instead of using a travel threshold.

[joseph_g3]

@Craig DIckson

Another option would be to use a divider layout like this:

As long as you don't include the queue as a barrier (as long as the queue isn't connected to the A* navigator), the AGVs should travel right up to the queue when unloading items.

[craig_dickson]

Hmm. The problem with what you drew is that I need the harbor to be 2 vehicles wide, and after the Unload task the vehicle is directed to move aside into the adjacent parking location, and that location happens to contain another red point (it's the top right in your picture). So sometimes a vehicle comes to the parking location to drop instead.

The other problem with the Travel to Loc vs Unload is that I use the Unload to trigger that movement to parking. Using a Travel to Loc and then a Move item doesn't fire that trigger. (See my next question posted about half an hour ago.)

[philboboADSK]

Just use a Travel to Loc and then an Unload.

It doesn't have to be one or the other.

[craig_dickson]

@phil.bobo I did try that but got some very odd behavior in AStar. I posted another question with a model that show what happens. Thanks

[joseph_g3]

@Craig DIckson

If you're going to be using Process Flow anyway, you could just use Travel to Loc to send the task executor to the unload point, move the item off, and then use a Travel to Loc or just plain Travel to move it into the parking spot. Sure, the parking spot would be different depending on which unload location the task executor goes to, but you could always use custom code or a table to determine where to park.

[philboboADSK]

That model doesn't have any Travel To Loc tasks in it, just Travel tasks.

Use a Travel To Loc task:

[craig_dickson]

Yes, that's the direction I am headed. That will probably work for the application.

Still, I'd like to know why using "Unload" in AStar sent the TE on an apparent wild goose chase though. If you connect to either the left or right code blocks you'll see what I mean. In previous experience both of those work fine outside of AStar?

[joseph_g3]

@Craig DIckson

See my answer to Raul's question: A Star not behaving properly?