vtpp2014 / Hybrid-Artificial-Potential-Field-A-star-Planning

This method takes best of both world. On one hand,it tries to reduce the overall path cost by using A-star and on other hand it reduces the time complexity by adapting real time reactive power from Artificial-Potential method of motion Planning.

Hybrid-Artificial-Potential-Field-A-star-Planning

This method takes best of both world. On one hand,it tries to reduce the overall path cost by using A-star and on other hand it reduces the time complexity by adapting real time reactive power from Artificial-Potential method of motion Planning. Welcome to the Hybrid-Artificial-Potential-Field-A-star-Planning wiki!

Here we are trying to mix two known method of motion planning , A-star and Artificial-Potential Field method.

We are implementing these two methods using following two mixtures(Environment is considered to be static).

We have find the planned path using A*, which will run offline ad find a path from source to destination. Now two variations of reactive method (Artificial Potential field ) is boosted by this already planned path.

1. We know that A-star is offline algorithm, which calculates path to goal directly without any feedback from next frame. Now we can divide our planned path into local goals and use Artificial-potential method to get to those. This way it will be faster and cheaper.The algorithm goes like this.

local_goals = divide path from A*
local_source = current source
local_goal = local_goals[0]
path = empty_list
while local_source != final_goal:
    path += Artificial_potential_path(local_source, local_goal)
    local_source = local_goal
    local_goal = next(local_goals)
print path

2)Here we use A-star and Potential field simultaneously depending on some parameters. Currently in this code we are considering that parameter as distance from nearest obstacle. So algorithm goes like this.

if distance from nearest obstacle > k(some parameter):
    next position = A-star planner
else:
    next position = Artificial-Potential-field planner

---

Prerequisites

• Python
• OpenCV
• Numpy
• Matplotlib

Input Images It will take all images in root folder as input images.

Sample Input Images:

![alt text][logo1] [logo1]: 1.jpg "Sample Image"

![alt text][logo2] [logo2]: 2.jpg "Sample Image"

![alt text][logo3] [logo3]: 3.jpg "Sample Image"

Output Type 1:

![alt text][logo4] [logo4]: output1/1.jpg "Sample Image"

![alt text][logo5] [logo5]: output1/2.jpg "Sample Image"

![alt text][logo6] [logo6]: output1/3.jpg "Sample Image"

Output Type 2:

![alt text][logo7] [logo7]: output2/1.jpg "Sample Image"

![alt text][logo8] [logo8]: output2/2.jpg "Sample Image"

![alt text][logo9] [logo9]: output2/3.jpg "Sample Image"