c# - A* pathfinder obstacle collision problem -

I am working on a project with a robot for which the object has to find a path and some obstacles The objects that are far away are for taking it.

The problem is that robots and objects need to take robots, both pixeters are one pixel wide in fact they are really big, often A * Pathfinder chooses to choose the route with the edges of the obstacles Sometimes, it collides with them, which we do not want to do.

I have some other non-barriers to run, but it does not always work very well, it still collides with obstacles, even adding several points where it is moving There is no way in it as a result, there is no way that can run.

Do you have any suggestions about this problem?

Edit:

So I suggested Justin L by adding a lot of costs around the obstacles, as a result:

Here you can barriers Look at the cost around, in the middle, two obstacles should look like people in the corner, but after running our illusion it seems that if cost override:

The picture shows what things get on the picture.

This is the way, which was also our problem before, which was hugging obstacles.

And another picture with the cost map on the way.

So what makes me feel strange that A * Pathshield is overriding these area costs, which is very high.

Does this open list evaluate the nodes inside intelligence to see the current field, whether the current field path is less than one in the open list?

And here is the code that I am using for Pathfinder:

Pathfinder.cs:

Field.cs and Grid.cs:

Do you have a shield costing pixels near one object?

Perhaps a linear shield is as simple:

  c = -mx + b  

where x is for the closest object The distance is B, the cost is out of bounds, and the meter is the rate at which the value of the cost is eliminated. Of course, if C is negative, then it should be set to 0.

Perhaps a simple hyperbolic decay

  c = b / x  

where B is the desired cost outside of the right threshold, again <0 p>

Alternatively, you can use exponential decay.

  C = K ^ (- Hx)  

< P> Where k scaling is stable, and h is the rate of precipitation. Again, cut off is smart.

---

Second Tip

I have never applied a * on a pixel-maped map; Almost always, tile

Can you try to reduce the "resolution" of your tiles in a big way? Probably a set of ten-ten-or twenty-twenty-pixels per one tile; Tile cost is the highest cost of a pixel in the tile.

In addition, you can try to de-value the shortest approximation used for A *.