Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

added FrozenLake environment #135

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

added FrozenLake environment #135

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
138 changes: 138 additions & 0 deletions src/envs/FrozenLakeEnv.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,138 @@

export FrozenLakeEnv

Base.@kwdef mutable struct frozenlakeEnv <: AbstractEnv
# lake: 3 is the spawn point, 1 is the goal and 2 are obstacles.
lake::Array{Int8,2} = [1 0 0 0 0
0 2 0 2 0
0 0 0 2 0
0 2 0 0 0
0 0 0 0 3]
reward::Union{Nothing, Float64} = 0
observation::Int = 24
is_terminated = false
num_steps::Int8 = 0
MAX_STEPS::Int8 = 100
end

RLBase.action_space(env::frozenlakeEnv) = (:up, :right, :down, :left)

begin
RLBase.state(env::frozenlakeEnv) = env.observation
RLBase.state_space(env::frozenlakeEnv) = collect(0:24)
RLBase.is_terminated(env::frozenlakeEnv) = env.is_terminated
end

function RLBase.reset!(env::frozenlakeEnv)
env.reward = 0
env.observation = 24
env.num_steps = 0
env.is_terminated = false
end

function moveDeterministic(action::String, observation)
if action == "up"
observation = clamp((observation÷5)-1, 0, 4)*5 + (observation - 5*(observation÷5))
elseif action == "down"
observation = clamp((observation÷5)+1, 0, 4)*5 + (observation - 5*(observation÷5))
elseif action == "left"
observation = observation÷5*5 + clamp((observation - 5*(observation÷5))-1, 0, 4)
elseif action == "right"
observation = observation÷5*5 + clamp((observation - 5*(observation÷5))+1, 0, 4)
end
return observation
end


function (env::frozenlakeEnv)(action)
r = rand()
env.num_steps += 1
if action == :up
if r < 1/3
# goes up
env.observation = moveDeterministic("up", env.observation)
elseif r < 2/3
# goes left
env.observation = moveDeterministic("left", env.observation)
else
# goes right
env.observation = moveDeterministic("left", env.observation)
end

elseif action == :right
if r < 1/3
# goes right
env.observation = moveDeterministic("right", env.observation)
elseif r < 2/3
# goes up
env.observation = moveDeterministic("up", env.observation)
else
# goes down
env.observation = moveDeterministic("down", env.observation)
end

elseif action == :down
if r < 1/3
# goes right
env.observation = moveDeterministic("right", env.observation)
elseif r < 2/3
# goes left
env.observation = moveDeterministic("left", env.observation)
else
# goes down
env.observation = moveDeterministic("down", env.observation)
end

elseif action == :left
if r < 1/3
# goes up
env.observation = moveDeterministic("up", env.observation)
elseif r < 2/3
# goes left
env.observation = moveDeterministic("left", env.observation)
else
# goes down
env.observation = moveDeterministic("down", env.observation)
end
else
@error "unknown action of $action"
end

obdiv5 = env.observation÷5
if env.num_steps>env.MAX_STEPS
env.is_terminated = true
end
if env.lake[obdiv5+1, env.observation - (5*obdiv5) + 1] == 0 || env.lake[obdiv5+1, env.observation - (5*obdiv5) + 1] == 3
env.reward-=0.01
elseif env.lake[obdiv5+1, env.observation - (5*obdiv5) + 1] == 2
env.is_terminated = true
env.reward-=1
elseif env.lake[obdiv5+1, env.observation - (5*obdiv5) + 1] == 1
env.is_terminated = true
env.reward+=5
end
# println(env.reward," ", env.num_steps, " ", env.observation)
end

RLBase.reward(env::frozenlakeEnv) = env.reward

# Env = frozenlakeEnv()
# RLBase.test_runnable!(Env)
# begin
# run(RandomPolicy(action_space(Env)), Env, StopAfterEpisode(1))
# end

# begin
# hook = TotalRewardPerEpisode()
# run(RandomPolicy(action_space(Env)), Env, StopAfterEpisode(10), hook)
# plot(hook.rewards)
# end
# p = QBasedPolicy(
# learner = MonteCarloLearner(;
# approximator=TabularQApproximator(
# n_state = length(state_space(Env)),
# n_action = length(action_space(Env)),
# )
# ),
# explorer = EpsilonGreedyExplorer(0.1)
# )