From 2d054dc271fbec21f2455f1cb3641d1f76af615a Mon Sep 17 00:00:00 2001
From: Armin <armin.co@hs-bochum.de>
Date: Fri, 12 Feb 2021 20:35:15 +0100
Subject: [PATCH] Environment works with keras
---
carla_environment.py | 111 ++++++++++++++++++++++++++++++++++---------
main.py | 16 ++++---
steering_wheel.py | 56 +++++++++++++++++++++-
3 files changed, 152 insertions(+), 31 deletions(-)
diff --git a/carla_environment.py b/carla_environment.py
index 2eeb376..d1c660d 100644
--- a/carla_environment.py
+++ b/carla_environment.py
@@ -8,8 +8,9 @@ import weakref
import numpy as np
import pygame
+import time
-from steering_wheel import Controller
+from steering_wheel import ACTION_SPACE, Controller
# find carla module
try:
@@ -68,6 +69,7 @@ class Camera:
class CollisionSensor:
sensor = None
+ collision = None
def __init__(self, world, parent):
bp = world.get_blueprint_library().find('sensor.other.collision')
@@ -80,6 +82,7 @@ class CollisionSensor:
if not self:
return
print(event.other_actor)
+ self.collision = event
impulse = event.normal_impulse
intensity = math.sqrt(impulse.x**2 + impulse.y**2 + impulse.z**2)
print(intensity)
@@ -94,6 +97,7 @@ class World:
blueprint_library = None
spawn_points = None
actors = []
+ was_big_reward = False
def __init__(self, world):
self.world = world
@@ -104,9 +108,17 @@ class World:
def reset(self):
""" Remove and create new player/vehicle. """
self.destroy()
+ time.sleep(0.5)
self.spawn_player()
self.spawn_actors()
+ def observation(self):
+ if self.player is not None:
+ pos = self.player.get_transform()
+ return [pos.location.x, pos.location.y, pos.rotation.yaw]
+ else:
+ return [0,0,0]
+
def spawn_player(self):
""" Add a vehicle to the world. """
while self.player is None:
@@ -125,10 +137,11 @@ class World:
transforms = []
transforms.append(carla.Transform(carla.Location(x=-19.2, y=54.5, z=0.2), carla.Rotation(pitch=0.0, yaw=-180, roll=0.0)))
transforms.append(carla.Transform(carla.Location(x=-14.3, y=54.7, z=0.2), carla.Rotation(pitch=0.0, yaw=-180, roll=0.0)))
- transforms.append(carla.Transform(carla.Location(x=-41.3, y=53.7, z=0.2), carla.Rotation(pitch=0.0, yaw=-180, roll=0.0)))
-
+ transforms.append(carla.Transform(carla.Location(x=-37.3, y=53.7, z=0.2), carla.Rotation(pitch=0.0, yaw=-180, roll=0.0)))
+ mass = carla.VehiclePhysicsControl(mass=900000.0)
for transform in transforms:
actor = self.world.try_spawn_actor(blueprint, transform)
+ actor.apply_physics_control(mass)
if actor is not None:
self.actors.append(actor)
@@ -136,48 +149,102 @@ class World:
""" Remove vehicle from the world. """
if self.player is not None:
self.player.destroy()
+ self.player = None
if self.actors is not None:
for actor in self.actors:
actor.destroy()
+ self.actors = []
def step(self, action):
""" Apply controls to vehicle. """
- self.player.apply_control(action)
- # print(str(self.player.get_transform()))
- # print(str(self.player.get_velocity()))
-
-
+ controls = Controller.action_to_controls(action)
+ c = carla.VehicleControl(throttle=controls[0], steer=controls[1], brake=controls[2], reverse=controls[3])
+ self.player.apply_control(c)
+ reward = self.reward()
+ self.collision_sensor.collision = None
+ return reward
+
+ def reward(self):
+ target = carla.Transform(carla.Location(x=-25.534311, y=54.460903, z=0.112781), carla.Rotation(pitch=0.000000, yaw=-175.922913, roll=-6.221135))
+ pos = self.player.get_transform()
+ pos_diff = (target.location.x - pos.location.x)**2 + (target.location.y - pos.location.y)**2 + 1
+ rot_diff = (target.rotation.yaw - pos.rotation.yaw)**2 + 1
+ r = ((1/pos_diff) - 0.5)
+ at_final_position = False
+ if pos_diff < 1.2:
+ if self.was_big_reward == False:
+ r += 100
+ self.was_big_reward = True
+ if rot_diff < 5:
+ at_final_position = True
+ else:
+ if self.was_big_reward == True:
+ self.was_big_reward = False
+ r -= 50
+
+ velocity = self.player.get_velocity()
+ v = (velocity.x)**2 + (velocity.y)**2
+ done = False
+ if at_final_position and v < 0.01:
+ done = True
+ if self.collision_sensor.collision is not None:
+ r -= 100
+ done = True
+ return self.observation(), r, done, None
+
+class ActionSpace:
+ n = ACTION_SPACE
+
+class ObservationSpace:
+ shape = [3]
class CarlaEnvironment:
+ action_space = ActionSpace
+ observation_space = ObservationSpace
world = None
client = None
+ camera = None
+ allow_render = False
- def __init__(self, host="127.0.0.1", port=2000):
+ def __init__(self, host="127.0.0.1", port=2000, render=False):
+ pygame.init()
self.client = carla.Client(host, port)
self.client.set_timeout(2.0)
self.client.load_world('Town07')
self.world = World(self.client.get_world())
+ if render:
+ self.allow_render = True
+ self.camera = Camera(self.world, camera_type='semantic_segmentation')
+ print(self.observation_space.shape[0])
- def step(self, action):
- control = carla.VehicleControl(throttle=action[0], steer=action[2], brake=action[1], reverse=action[3])
- self.world.step(control)
+ def reset(self):
+ self.world.reset()
+ return self.world.observation()
+
+ def render(self):
+ if self.allow_render:
+ self.camera.on_update()
+
+ def step(self, action, render=False):
+ if render:
+ self.render()
+ return self.world.step(action)
if __name__ == "__main__":
LOCALHOST="127.0.0.1"
SERVER="192.168.188.20"
-
- pygame.init()
+ env = CarlaEnvironment(host=LOCALHOST, render=True)
clock = pygame.time.Clock()
-
- env = CarlaEnvironment(host=LOCALHOST)
- cam = Camera(env.world, camera_type='semantic_segmentation')
ctrl = Controller()
-
+ cumulated_reward = 0
while ctrl.is_running():
- clock.tick(60)
+ clock.tick(20)
ctrl.on_update()
- env.step(ctrl.get_action())
- cam.on_update()
-
+ obs, reward, done = env.step(ctrl.get_action(), render=True)
+ if done:
+ break
+ print(str(reward) + ' ' + str(done))
+ cumulated_reward += reward
env.world.destroy()
pygame.quit()
+ print(cumulated_reward)
diff --git a/main.py b/main.py
index 254f2b6..f09752b 100644
--- a/main.py
+++ b/main.py
@@ -4,9 +4,10 @@ Run your desired environment and agent configuration.
import os
import atexit
-import gym
+from carla_environment import CarlaEnvironment
+# import gym
-from agents import DQAgent as QAgent
+from agents import QAgent as QAgent
import environment_wrapper as ew
# Allow GPU usage or force tensorflow to use the CPU.
@@ -15,13 +16,15 @@ if FORCE_CPU:
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"] = ""
-
+RENDER = False
if __name__ == '__main__':
# 1. Create an environment
- env = gym.make('LunarLander-v2')
+ env = CarlaEnvironment(render=RENDER)
+ # env = gym.make('LunarLander-v2')
+ print(env.observation_space.shape[0])
# 2. Create a learning agent
- marvin = QAgent(env.action_space.n, env.observation_space.shape[0], 'FromScratchDouble')
+ marvin = QAgent(env.action_space.n, env.observation_space.shape[0], 'CarlaTest')
# (2.5) *optional* Load agent memory and/or net from disk.
agnt = 'agent'
@@ -31,11 +34,10 @@ if __name__ == '__main__':
marvin.load('saved_agents/' + agnt + '/' + agnt, net=LOAD_ANN, memory=LOAD_MEMORIES)
# 3. Set your configurations for the run.
- RENDER = False
LEARNING = True
LEARN_ONLINE = True
LEARN_OFFLINE = False
- RUN_EPISODES = 500
+ RUN_EPISODES = 10
LEARN_OFFLINE_EPOCHS = 500
SAVE_PATH = "./saved_agents"
diff --git a/steering_wheel.py b/steering_wheel.py
index 66e031e..2828f7b 100644
--- a/steering_wheel.py
+++ b/steering_wheel.py
@@ -9,6 +9,18 @@ JOYSTICK_REVERSE=3
INPUT_OFFSET = 2.0
+IDLE=0
+FORWARD=1
+REVERSE=2
+FORWARD_LEFT=3
+FORWARD_RIGHT=4
+LEFT=5
+RIGHT=6
+BRAKE=7
+REVERSE_LEFT=8
+REVERSE_RIGHT=9
+ACTION_SPACE=10
+
class ManualSteeringWheel:
""" Steering wheel """
axis_count = 0.0
@@ -52,9 +64,49 @@ class ManualSteeringWheel:
self.get_throttle(update=False)
self.get_brakes(update=False)
+ def action_to_controls(action):
+ if action == IDLE:
+ return [0.0, 0.0, 0.0, False]
+ if action == FORWARD:
+ return [0.8, 0.0, 0.0, False]
+ if action == REVERSE:
+ return [0.8, 0.0, 0.0, True]
+ if action == FORWARD_LEFT:
+ return [0.8, -0.4, 0.0, False]
+ if action == FORWARD_RIGHT:
+ return [0.8, 0.4, 0.0, False]
+ if action == LEFT:
+ return [0.0, -0.4, 0.0, False]
+ if action == RIGHT:
+ return [0.0, 0.5, 0.0, False]
+ if action == BRAKE:
+ return [0.0, 0.0, 0.5, False]
+ if action == REVERSE_LEFT:
+ return [0.5, -0.4, 0.0, True]
+ if action == REVERSE_RIGHT:
+ return [0.5, 0.4, 0.0, True]
+
def get_action(self):
-
- return [self.throttle, self.brakes, self.direction, self.reverse]
+ action = IDLE
+ if self.throttle > 0.5 and self.brakes < 0.5 and abs(self.direction) < 0.1 and not self.reverse:
+ action = FORWARD
+ elif self.throttle > 0.5 and self.brakes < 0.5 and self.direction < -0.2 and not self.reverse:
+ action = FORWARD_LEFT
+ elif self.throttle > 0.5 and self.brakes < 0.5 and self.direction > 0.2 and not self.reverse:
+ action = FORWARD_RIGHT
+ elif self.throttle > 0.5 and self.brakes < 0.5 and abs(self.direction) < 0.2 and self.reverse:
+ action = REVERSE
+ elif self.throttle < 0.5 and self.brakes < 0.5 and self.direction < -0.2:
+ action = LEFT
+ elif self.throttle < 0.5 and self.brakes < 0.5 and self.direction > 0.2:
+ action = RIGHT
+ elif self.brakes > 0.5:
+ action = BRAKE
+ elif self.throttle > 0.5 and self.brakes < 0.5 and self.direction < -0.2 and self.reverse:
+ action = REVERSE_LEFT
+ elif self.throttle > 0.5 and self.brakes < 0.5 and self.direction > 0.2 and self.reverse:
+ action = REVERSE_RIGHT
+ return action
class Controller(ManualSteeringWheel):
--
GitLab