Copy folder custom vao thu muc "gym_gazebo/envs"
Copy folder launch, models, worlds vao thu muc "gym_gazebo/envs/assets". skip nhung file trung
Copy file init.py vao trong thu muc "gym_gazebo" (copy de)
vao trong thu muc "gym_gazebo/envs/assets/launch", mo file MazeColor.launch, sua truong defaul o "" tro toi file "maze_color.world" trong folder "worlds"
Vao trong folder "ForLearn", chay lenh terminal "python DeepQlearningWithEgreedy.py"
Target la hinh tru do
Hint la cac vong tron nhieu mau
Nhiem vu la cho turtlebot hoc cach di chuyen den target
file dinh nghia moi truong
+ hien tai dang tra ve observation la 3 gia tri: anh mau RGB(32*32*3), laserscaner, va position cua turtlebot
+ reward: 200 khi cham den dich, -200 khi dam vao tuong, 10 khi cham lan dau tien vao hint
+ action: gom 5 action: 2 re trai, 1 di thang, 2 re phai
file training: dang cai dat thuat toan Q-learning voi E-greedy
class chua cac experiences cua turtlebot
class Qnetwork:
thong tin ve mang neural network
co mot so ham ho tro
Viec thay doi cau truc mang neural network dc thay doi trong file nay
cac class ho tro viec luu du lieu va luu cac thong so
Co the su dung 1 hoac ca 3 du lieu dua ra tu environment
viec thay doi cau truc mang, thuat toan dc sua trong file Qnetwork.py
viec thay doi strategy chon action dc thay doi trong file DeepQlearing...
viec thay doi cac du lieu moi truong trong file gazebo_turtlebot_maze_color.py
thay doi thong so trong file Ultility.py
###Hien tai co 3 target trong moi truong duoc random trong moi lan chay
cac hint duoc tinh diem tai lan dau tien cham phai