Self Driving Car - Python Plays GTA V
As I was contemplating the next steps, I was curious about a couple of choices for Artificially Intelligent driving. One thought I had was that we could detect both lanes, and then attempt to orient the car in between the two lanes, so long as those two lanes had different slopes.
...then I began to think about it, and wondered if the problem coudl be even more simple than that. I noticed that the times when the lanes would "both" be on one side or the other was when we were getting to close to one of the edges.
So next I wondered, hmm, what if we just work with the following logic:
If one lane's slope is positive, and the other is negative, then we're fine, continue straight.
If both lane's slopes are negative, then we're too far left, and we should turn right.
If both lane's slopes are posotive, then we're too far right, and we should turn left.
...surely it can't be that simple right?
Starting code:
import numpy as np
from PIL import ImageGrab
import cv2
import time
from numpy import ones,vstack
from numpy.linalg import lstsq
from directkeys import PressKey, W, A, S, D
from statistics import mean
def roi(img, vertices):
#blank mask:
mask = np.zeros_like(img)
#filling pixels inside the polygon defined by "vertices" with the fill color
cv2.fillPoly(mask, vertices, 255)
#returning the image only where mask pixels are nonzero
masked = cv2.bitwise_and(img, mask)
return masked
def draw_lanes(img, lines, color=[0, 255, 255], thickness=3):
# if this fails, go with some default line
try:
# finds the maximum y value for a lane marker
# (since we cannot assume the horizon will always be at the same point.)
ys = []
for i in lines:
for ii in i:
ys += [ii[1],ii[3]]
min_y = min(ys)
max_y = 600
new_lines = []
line_dict = {}
for idx,i in enumerate(lines):
for xyxy in i:
# These four lines:
# modified from http://stackoverflow.com/questions/21565994/method-to-return-the-equation-of-a-straight-line-given-two-points
# Used to calculate the definition of a line, given two sets of coords.
x_coords = (xyxy[0],xyxy[2])
y_coords = (xyxy[1],xyxy[3])
A = vstack([x_coords,ones(len(x_coords))]).T
m, b = lstsq(A, y_coords)[0]
# Calculating our new, and improved, xs
x1 = (min_y-b) / m
x2 = (max_y-b) / m
line_dict[idx] = [m,b,[int(x1), min_y, int(x2), max_y]]
new_lines.append([int(x1), min_y, int(x2), max_y])
final_lanes = {}
for idx in line_dict:
final_lanes_copy = final_lanes.copy()
m = line_dict[idx][0]
b = line_dict[idx][1]
line = line_dict[idx][2]
if len(final_lanes) == 0:
final_lanes[m] = [ [m,b,line] ]
else:
found_copy = False
for other_ms in final_lanes_copy:
if not found_copy:
if abs(other_ms*1.2) > abs(m) > abs(other_ms*0.8):
if abs(final_lanes_copy[other_ms][0][1]*1.2) > abs(b) > abs(final_lanes_copy[other_ms][0][1]*0.8):
final_lanes[other_ms].append([m,b,line])
found_copy = True
break
else:
final_lanes[m] = [ [m,b,line] ]
line_counter = {}
for lanes in final_lanes:
line_counter[lanes] = len(final_lanes[lanes])
top_lanes = sorted(line_counter.items(), key=lambda item: item[1])[::-1][:2]
lane1_id = top_lanes[0][0]
lane2_id = top_lanes[1][0]
def average_lane(lane_data):
x1s = []
y1s = []
x2s = []
y2s = []
for data in lane_data:
x1s.append(data[2][0])
y1s.append(data[2][1])
x2s.append(data[2][2])
y2s.append(data[2][3])
return int(mean(x1s)), int(mean(y1s)), int(mean(x2s)), int(mean(y2s))
l1_x1, l1_y1, l1_x2, l1_y2 = average_lane(final_lanes[lane1_id])
l2_x1, l2_y1, l2_x2, l2_y2 = average_lane(final_lanes[lane2_id])
return [l1_x1, l1_y1, l1_x2, l1_y2], [l2_x1, l2_y1, l2_x2, l2_y2]
except Exception as e:
print(str(e))
def process_img(image):
original_image = image
# convert to gray
processed_img = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# edge detection
processed_img = cv2.Canny(processed_img, threshold1 = 200, threshold2=300)
processed_img = cv2.GaussianBlur(processed_img,(5,5),0)
vertices = np.array([[10,500],[10,300],[300,200],[500,200],[800,300],[800,500],
], np.int32)
processed_img = roi(processed_img, [vertices])
# more info: http://docs.opencv.org/3.0-beta/doc/py_tutorials/py_imgproc/py_houghlines/py_houghlines.html
# rho theta thresh min length, max gap:
lines = cv2.HoughLinesP(processed_img, 1, np.pi/180, 180, 20, 15)
try:
l1, l2 = draw_lanes(original_image,lines)
cv2.line(original_image, (l1[0], l1[1]), (l1[2], l1[3]), [0,255,0], 30)
cv2.line(original_image, (l2[0], l2[1]), (l2[2], l2[3]), [0,255,0], 30)
except Exception as e:
print(str(e))
pass
try:
for coords in lines:
coords = coords[0]
try:
cv2.line(processed_img, (coords[0], coords[1]), (coords[2], coords[3]), [255,0,0], 3)
except Exception as e:
print(str(e))
except Exception as e:
pass
return processed_img,original_image
def main():
last_time = time.time()
while True:
screen = np.array(ImageGrab.grab(bbox=(0,40,800,640)))
print('Frame took {} seconds'.format(time.time()-last_time))
last_time = time.time()
new_screen,original_image = process_img(screen)
cv2.imshow('window', new_screen)
cv2.imshow('window2',cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB))
#cv2.imshow('window',cv2.cvtColor(screen, cv2.COLOR_BGR2RGB))
if cv2.waitKey(25) & 0xFF == ord('q'):
cv2.destroyAllWindows()
break
First, we need to log the slopes of the top two lines. From our draw_lanes function, we can use lane1_id and lane2_id, since these were both slopes for clusters of lines, and these are the top 2 slopes by line count of similar slopes. So, at the end of the draw_lanes function, we just need to also return those two new things:
def draw_lanes(img, lines, color=[0, 255, 255], thickness=3):
# if this fails, go with some default line
try:
# finds the maximum y value for a lane marker
# (since we cannot assume the horizon will always be at the same point.)
ys = []
for i in lines:
for ii in i:
ys += [ii[1],ii[3]]
min_y = min(ys)
max_y = 600
new_lines = []
line_dict = {}
for idx,i in enumerate(lines):
for xyxy in i:
# These four lines:
# modified from http://stackoverflow.com/questions/21565994/method-to-return-the-equation-of-a-straight-line-given-two-points
# Used to calculate the definition of a line, given two sets of coords.
x_coords = (xyxy[0],xyxy[2])
y_coords = (xyxy[1],xyxy[3])
A = vstack([x_coords,ones(len(x_coords))]).T
m, b = lstsq(A, y_coords)[0]
# Calculating our new, and improved, xs
x1 = (min_y-b) / m
x2 = (max_y-b) / m
line_dict[idx] = [m,b,[int(x1), min_y, int(x2), max_y]]
new_lines.append([int(x1), min_y, int(x2), max_y])
final_lanes = {}
for idx in line_dict:
final_lanes_copy = final_lanes.copy()
m = line_dict[idx][0]
b = line_dict[idx][1]
line = line_dict[idx][2]
if len(final_lanes) == 0:
final_lanes[m] = [ [m,b,line] ]
else:
found_copy = False
for other_ms in final_lanes_copy:
if not found_copy:
if abs(other_ms*1.2) > abs(m) > abs(other_ms*0.8):
if abs(final_lanes_copy[other_ms][0][1]*1.2) > abs(b) > abs(final_lanes_copy[other_ms][0][1]*0.8):
final_lanes[other_ms].append([m,b,line])
found_copy = True
break
else:
final_lanes[m] = [ [m,b,line] ]
line_counter = {}
for lanes in final_lanes:
line_counter[lanes] = len(final_lanes[lanes])
top_lanes = sorted(line_counter.items(), key=lambda item: item[1])[::-1][:2]
lane1_id = top_lanes[0][0]
lane2_id = top_lanes[1][0]
def average_lane(lane_data):
x1s = []
y1s = []
x2s = []
y2s = []
for data in lane_data:
x1s.append(data[2][0])
y1s.append(data[2][1])
x2s.append(data[2][2])
y2s.append(data[2][3])
return int(mean(x1s)), int(mean(y1s)), int(mean(x2s)), int(mean(y2s))
l1_x1, l1_y1, l1_x2, l1_y2 = average_lane(final_lanes[lane1_id])
l2_x1, l2_y1, l2_x2, l2_y2 = average_lane(final_lanes[lane2_id])
return [l1_x1, l1_y1, l1_x2, l1_y2], [l2_x1, l2_y1, l2_x2, l2_y2], lane1_id, lane2_id
except Exception as e:
print(str(e))
Next, we need to handle for this in our process_img function, so when we call the draw_lanes, we also store the slopes:
def process_img(image):
original_image = image
# convert to gray
processed_img = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# edge detection
processed_img = cv2.Canny(processed_img, threshold1 = 200, threshold2=300)
processed_img = cv2.GaussianBlur(processed_img,(5,5),0)
vertices = np.array([[10,500],[10,300],[300,200],[500,200],[800,300],[800,500],
], np.int32)
processed_img = roi(processed_img, [vertices])
# more info: http://docs.opencv.org/3.0-beta/doc/py_tutorials/py_imgproc/py_houghlines/py_houghlines.html
# rho theta thresh min length, max gap:
lines = cv2.HoughLinesP(processed_img, 1, np.pi/180, 180, 20, 15)
m1 = 0
m2 = 0
try:
l1, l2, m1,m2 = draw_lanes(original_image,lines)
cv2.line(original_image, (l1[0], l1[1]), (l1[2], l1[3]), [0,255,0], 30)
cv2.line(original_image, (l2[0], l2[1]), (l2[2], l2[3]), [0,255,0], 30)
except Exception as e:
print(str(e))
pass
try:
for coords in lines:
coords = coords[0]
try:
cv2.line(processed_img, (coords[0], coords[1]), (coords[2], coords[3]), [255,0,0], 3)
except Exception as e:
print(str(e))
except Exception as e:
pass
return processed_img,original_image, m1, m2
Note that we also define both slopes to be a default of 0, just in case our draw_lanes fails, which it can. Now, let's define some driving functions:
def straight():
PressKey(W)
ReleaseKey(A)
ReleaseKey(D)
def left():
PressKey(A)
ReleaseKey(W)
ReleaseKey(D)
ReleaseKey(A)
def right():
PressKey(D)
ReleaseKey(A)
ReleaseKey(W)
ReleaseKey(D)
def slow_ya_roll():
ReleaseKey(W)
ReleaseKey(A)
ReleaseKey(D)
Finally:
def main():
last_time = time.time()
while True:
screen = np.array(ImageGrab.grab(bbox=(0,40,800,640)))
print('Frame took {} seconds'.format(time.time()-last_time))
last_time = time.time()
new_screen,original_image, m1, m2 = process_img(screen)
#cv2.imshow('window', new_screen)
cv2.imshow('window2',cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB))
if m1 < 0 and m2 < 0:
right()
elif m1 > 0 and m2 > 0:
left()
else:
straight()
#cv2.imshow('window',cv2.cvtColor(screen, cv2.COLOR_BGR2RGB))
if cv2.waitKey(25) & 0xFF == ord('q'):
cv2.destroyAllWindows()
break
It's not perfect, and sometimes it gets going too fast and runs out of the lane and gets lost. What's intersting to me, however, is that this actually works a good deal of the time. When it doesn't work, the lane finders goes nuts and this is easy for us to determine what the situation is.
For me, that's a big deal, because now we can begin to build some training data, where the determinant of success is mainly when we're dealing with a positive and negative slope situation, OR we go quickly from a same sign slope situation back to a different signed slope situation (correctly correcting a mistake).
From here, we can use this to build a training dataset, and even try more scenarios.
-
Reading game frames in Python with OpenCV - Python Plays GTA V
-
OpenCV basics - Python Plays GTA V
-
Direct Input to Game - Python Plays GTA V
-
Region of Interest for finding lanes - Python Plays GTA V
-
Hough Lines - Python Plays GTA V
-
Finding Lanes for our self driving car - Python Plays GTA V
-
Self Driving Car - Python Plays GTA V
-
Next steps for Deep Learning self driving car - Python Plays GTA V
-
Training data for self driving car neural network- Python Plays GTA V
-
Balancing neural network training data- Python Plays GTA V
-
Training Self-Driving Car neural network- Python Plays GTA V
-
Testing self-driving car neural network- Python Plays GTA V
-
A more interesting self-driving AI - Python Plays GTA V
-
Object detection with Tensorflow - Self Driving Cars in GTA
-
Determining other vehicle distances and collision warning - Self Driving Cars in GTA
-
Getting the Agent a Vehicle- Python Plays GTA V
-
Acquiring a Vehicle for the Agent - Python Plays GTA V
