Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/jafarabdi/robond_first_project


https://github.com/jafarabdi/robond_first_project

Last synced: 2 days ago
JSON representation

Awesome Lists containing this project

README 

        
# RoboND_First_Project

---------------------------------------------



Notebook Analysis



1- to identify the rocks I writed a function called detect_rock() I converted the image to HSV format after that run inRange function to detect the color that I specify (Here yellow) I found the color range by runing the code in the end of this page http://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_imgproc/py_colorspaces/py_colorspaces.html#how-to-find-hsv-values-to-track



in this image it shown how the function detected the rock and it gives a good result



![Detect Rocks](https://github.com/JafarAbdi/RoboND_First_Project/blob/master/detect_rocks.png?raw=true)



---------------------------------------------



2- to detect the obstacles I had two options 



   1- by calling cv2.bitwise_not which will select the color that are not terrain 



   2- by using http://docs.opencv.org/trunk/d9/d61/tutorial_py_morphological_ops.html function (this method was recommended by a fellow in slack channel)

   in my code I used the second method



the steps used in detect obstacles function are as follow 



   1- detect the terrain area (left up corner image)



   2- dilating the terrain area with one iteration (right up corner)



   3- applying bitwise_not function to the original terrian (left down corner)



   4- taking bitwise_and between the dilated terrain and terrian not (right down corner)



![Detect Obstacles](https://github.com/JafarAbdi/RoboND_First_Project/blob/master/detect_obstacle.png?raw=true)



---------------------------------------------



1- for process_image function in notebook I did the same procedure for perception_step function in perception.py except rather than using Rover I used the Databucket instance data and the input image



Autonomous Navigation and Mapping



1- Video -> https://www.youtube.com/watch?v=gOOxzLtntbM

   FPS between 16-20

   Resultion 640x480



I added a detect_rock function and filled rotate_pix and translate_pix



For perception_step function 



-I difeined some constant bottom_offset, scale factor 10cm/m(pixle), size of destination square 2*5



-Add mask to limit the terrain and obstacles



-Define source and destination points



-Apply perspect_transform to the input image



-Identify the terrain and rocks, obstacles and mask them



-Add thersholded images to Rover.vision_image



-Convert map image pixel values to rover-centric coordinates after that to the world coordinates



-Update Rover.worldmap if the Rover.pitch <= 0.75 or Rover.pitch >= 359.25 and Rover.roll <= 1 or Rover.roll >= 359 the values selected experimentally



-Convert rover-centric pixel positions to polar coordinates by applying to_polar_coords to terrain pixles



I didn't do any change to decision_step function (I'll do when I start playing with Udacity challenge)



---------------------------------------------



I'm working now on udacity challenge 



to pick the samples I added 



if Rover.near_sample and not Rover.picking_up:



               # Set mode to "stop" and hit the brakes!



                Rover.throttle = 0

                

                # Set brake to stored brake value

                

                Rover.brake = Rover.brake_set

                

                Rover.steer = 0

                

                Rover.send_pickup = True

                

                Rover.samples_picked[np.argmin(Rover.dis_to_samples)] = 1



to decision_step function



and I'll add new variable to Rover class which store the magnitude of the distance from the rover to samples and pick them from min to max distance after that return to start position 



------------------------------------------------------



Just to mention in the previous submision the reviewer said that there's an error when running drive_rover.py I tried it multiple time in my computer without any errors :)