def load_new(cls, filename, load_from_package=None):
env_dict = cls.load_env_dict(filename,
load_from_package=load_from_package)
# TODO: inefficient - each one of these generators loads the complete env file.
env = rail_env.RailEnv(width=1,
height=1,
rail_generator=rail_gen.rail_from_file(
filename,
load_from_package=load_from_package),
schedule_generator=sched_gen.schedule_from_file(
filename,
load_from_package=load_from_package),
malfunction_generator_and_process_data=mal_gen.
malfunction_from_file(
filename,
load_from_package=load_from_package),
obs_builder_object=DummyObservationBuilder(),
record_steps=True)
env.rail = GridTransitionMap(1, 1) # dummy
cls.set_full_state(env, env_dict)
return env, env_dict
def test_build_graph(self):
manuel_graph = True
if manuel_graph:
specs = [[(8, 0), (1, 90), (1, 90), (6, 0), (1, 90), (7, 90)],
[(1, 180), (0, 0), (0, 0), (1, 0), (0, 0), (0, 0)],
[(8, 270), (1, 90), (1, 90), (2, 90), (1, 90), (7, 90)],
[(0, 0), (0, 0), (0, 0), (0, 0), (0, 0), (0, 0)],
[(1, 0), (0, 0), (0, 0), (0, 0), (0, 0), (0, 0)]]
rail_shape = np.array(specs).shape
env = rail_env.RailEnv(
width=rail_shape[1],
height=rail_shape[0],
rail_generator=rail_generators.
rail_from_manual_specifications_generator(specs),
number_of_agents=1)
else:
env = rail_env.RailEnv(
width=25,
height=15,
rail_generator=rail_generators.sparse_rail_generator())
observations, info = env.reset()
#env = rail_env.RailEnv(
# width=5,
# height=5,
# rail_generator=rail_generators.complex_rail_generator())
#obs, info = env.reset()
#obs, info = env.reset()
#obs, info = env.reset()
tg = transition_graph.TransitionGraph(env)
g = tg.g
plt = igraph.Plot()
layout = [(v['y'] + np.random.randn() * 0.15,
v['x'] + np.random.randn() * 0.15) for v in g.vs]
plt.add(g, layout=layout)
plt.redraw()
with tempfile.NamedTemporaryFile() as f:
plt.save(f.name)
graph_img = Image.open(f.name)
render = rendertools.RenderTool(env, gl='PILSVG')
render.render_env()
bg_img = Image.fromarray(render.get_image())
graph_img = graph_img.resize(bg_img.size)
minx, maxx, miny, maxy = min(g.vs['x']), max(g.vs['x']), min(
g.vs['y']), max(g.vs['y'])
envh, envw = env.height, env.width
w, h = bg_img.size
graph_img = graph_img.resize((int(
(maxy - miny) / envw * w), int((maxx - minx) / envh * h)))
bg_img = bg_img.resize(
(int(w / envw * (envw + 1)), int(h / envh * (envh + 1))))
bg_img.paste(graph_img,
(int(w / envw * (miny + .5)), int(h / envh *
(minx + .5))),
graph_img)
bg_img.show()
def test_build_graph(self):
manuel_graph = True
if manuel_graph:
specs = [[(8, 0), (1, 90), (1, 90), (6, 0), (1, 90), (8, 90),
(0, 0)],
[(1, 180), (0, 0), (0, 0), (1, 0), (0, 0), (1, 0),
(0, 0)],
[(8, 270), (1, 90), (1, 90), (2, 90), (1, 90), (2, 360),
(0, 0)],
[(0, 0), (0, 0), (0, 0), (0, 0), (0, 0), (1, 0), (0, 0)],
[(0, 0), (0, 0), (0, 0), (0, 0), (0, 0), (0, 0), (0, 0)]]
rail_shape = np.array(specs).shape
env = rail_env.RailEnv(
width=rail_shape[1],
height=rail_shape[0],
rail_generator=rail_generators.
rail_from_manual_specifications_generator(specs),
number_of_agents=1)
else:
env = rail_env.RailEnv(
width=25,
height=15,
rail_generator=rail_generators.sparse_rail_generator())
observations, info = env.reset()
tg = transition_graph.TransitionGraph(env)
g = tg.g
plt = igraph.Plot()
layout = [(v['y'] + np.random.randn() * 0.125,
v['x'] + np.random.randn() * 0.125) for v in g.vs]
plt.add(g,
layout=layout,
margin=50,
vertex_label=list(range(len(g.vs))),
edge_label=list(range(len(g.es))))
plt.redraw()
with tempfile.NamedTemporaryFile() as f:
plt.save(f.name)
graph_img = Image.open(f.name)
render = rendertools.RenderTool(env, gl='PILSVG')
render.render_env()
bg_img = Image.fromarray(render.get_image())
graph_img.show(
) #uncomment for higher res graph images to inspect labels etc.
minx, maxx, miny, maxy = min(g.vs['x']), max(g.vs['x']), min(
g.vs['y']), max(g.vs['y'])
envh, envw = env.height, env.width
w, h = bg_img.size
graph_img = graph_img.resize((int(
(maxy - miny) / envw * w), int((maxx - minx) / envh * h)))
bg_img = bg_img.resize(
(int(w / envw * (envw + 1)), int(h / envh * (envh + 1))))
bg_img.paste(graph_img,
(int(w / envw * (miny + .5)), int(h / envh *
(minx + .5))),
graph_img)
bg_img.show()
#import IPython
#IPython.embed()
print(
f"edges that share common resources: {transition_graph.find_edges_that_share_resource(g)}"
)
def test_build_graph(self):
manual_graph = False
if manual_graph:
specs = [[(8, 0), (1, 90), (1, 90), (6, 0), (1, 90), (8, 90),
(0, 0)],
[(1, 180), (0, 0), (0, 0), (1, 0), (0, 0), (1, 0),
(0, 0)],
[(8, 270), (1, 90), (1, 90), (2, 90), (1, 90), (2, 360),
(0, 0)],
[(0, 0), (0, 0), (0, 0), (0, 0), (0, 0), (1, 0), (0, 0)],
[(0, 0), (0, 0), (0, 0), (0, 0), (0, 0), (0, 0), (0, 0)]]
rail_shape = np.array(specs).shape
env = rail_env.RailEnv(
width=rail_shape[1],
height=rail_shape[0],
rail_generator=rail_generators.
rail_from_manual_specifications_generator(specs),
number_of_agents=1)
else:
env = rail_env.RailEnv(
width=25,
height=15,
rail_generator=rail_generators.sparse_rail_generator())
observations, info = env.reset()
def render_env():
render = rendertools.RenderTool(env, gl='PILSVG')
render.render_env()
bg_img = Image.fromarray(render.get_image()).convert('RGB')
img = ImageTk.PhotoImage(bg_img)
img.img = bg_img
return img
root = tk.Tk()
bg_img = render_env()
canvas = tk.Canvas(root,
width=bg_img.img.size[0],
height=bg_img.img.size[1])
canvas.pack()
canvas_img = canvas.create_image(0, 0, anchor=tk.NW, image=bg_img)
root.title('Yo')
def keypress(event):
nonlocal bg_img
k = event.keysym
action = dict(
Up=rail_env.RailEnvActions.MOVE_FORWARD,
Down=rail_env.RailEnvActions.STOP_MOVING,
Left=rail_env.RailEnvActions.MOVE_LEFT,
Right=rail_env.RailEnvActions.MOVE_RIGHT,
space=rail_env.RailEnvActions.DO_NOTHING,
)[k]
root.title('Loading...')
env.step({0: action})
bg_img = render_env()
canvas.itemconfig(canvas_img, image=bg_img)
root.title(f'Done: {k}')
for k in ('<Left>', '<Right>', '<Up>', '<Down>', '<space>'):
root.bind(k, keypress)
root.mainloop()