Utilize numpy to reduce calculation time #767
Conversation
Change-Id: I6e421a1c2524a3d8f8875121a1a6d2ed832c3150
Change-Id: I909d49b40e6792efcb796846c97a0d984471d3c9
|
The styling warnings are caused by W503, which says that it is going to change. |
| ) or ( | ||
| self.rhs[self.start.x, self.start.y] | ||
| != self.g[self.start.x, self.start.y] | ||
| ).any(): |
There was a problem hiding this comment.
Not this PR topic, but I think each while condition is challenging to read, so is it possible to introduce boolean variables or other kinds of intermediate values for readability?
There was a problem hiding this comment.
That is a great suggestion, thank you! I've updated this as you have suggested. Please take a look.
| if self.detected_obstacles_xy.shape[0] > 0: | ||
| is_in_detected_obstacles = ( | ||
| (self.detected_obstacles_xy[:, 0] == x) | ||
| & (self.detected_obstacles_xy[:, 1] == y) |
There was a problem hiding this comment.
Difficult to read each condition, is it possible to improve it?
There was a problem hiding this comment.
I separated the conditions with boolean variables.
| self.g = np.full( | ||
| (self.x_max, self.y_max), | ||
| float("inf") # type: ignore | ||
| ) | ||
| self.detected_obstacles = list() # type: ignore |
There was a problem hiding this comment.
self.dtected_obstacles is still needed?
There was a problem hiding this comment.
It is still used in random object generation. I will change that function as well and test it.
There was a problem hiding this comment.
I removed it, and while doing that fixed a bug about going out of range of the rhs.
| self.rhs = np.full( | ||
| (self.x_max, self.y_max), | ||
| float("inf") # type: ignore | ||
| ) |
There was a problem hiding this comment.
I think introducing a new function to initialize a grid map is better like:
def init2dGrid():
return np.full(self.x_max, self.y_max), float("inf")
There was a problem hiding this comment.
I used an already existing function called create_grid. Thanks for the improvement.
| @@ -52,6 +53,8 @@ class DStarLite: | |||
| Node(-1, -1, math.sqrt(2)) | |||
| ] | |||
|
|
|||
| initialized = False | |||
There was a problem hiding this comment.
This should be initizlied in the __init__?
There was a problem hiding this comment.
Yes, I moved it, thank you
Change-Id: Ic329c2140f2200b49feeb03c588d7a805b96cbdc
Change-Id: I4a0f2a424f17f44de3b444cb9c2e8e715d761d34
Change-Id: Id55c5972cbe76a2f6a69527dbf770f9bdf059109
Change-Id: I6e421a1c2524a3d8f8875121a1a6d2ed832c3150
Reference issue
What does this implement/fix?
I've noticed that the current Implementation uses lists. By replacing them with numpy arrays and using built in numpy functions such as any, I reduced the calculation time of d_star_lite. The other thing that I've noticed, once after initializing class, if the user wants to add new spoofed objects and calculate a new path, it was not using the prior information that is generated. So I modified the initialization process a bit to use the prior information.
Additional information
CheckList