Broken tests for tiger: convert_o ambiguity error

[BoZenKhaa]

The tests for tiger give an error because of the ambiguity of the convert_o method:

using POMDPModels
using POMDPTools
using POMDPs
using Random
using Test
using Compose

R = [-1. -100 10; -1 10 -100]

T = zeros(2,3,2)
T[:,:,1] = [1. 0.5 0.5; 0 0.5 0.5]
T[:,:,2] = [0. 0.5 0.5; 1 0.5 0.5]

O = zeros(2,3,2)
O[:,:,1] = [0.85 0.5 0.5; 0.15 0.5 0.5]
O[:,:,2] = [0.15 0.5 0.5; 0.85 0.5 0.5]

pomdp1 = TigerPOMDP()

pomdp2 = TabularPOMDP(T, R, O, 0.95)

policy = RandomPolicy(pomdp1, rng=MersenneTwister(2))
sim = RolloutSimulator(rng=MersenneTwister(3), max_steps=100)

simulate(sim, pomdp1, policy, updater(policy), initialstate(pomdp1))

o = last(observations(pomdp1))
@test o == 1
# test vec
ov = convert_o(Array{Float64}, true, pomdp1)
@test ov == [1.]
o = POMDPs.convert_o(Bool, ov, pomdp1)
@test o == true

returns

ERROR: MethodError: convert_o(::Type{Bool}, ::Vector{Float64}, ::TigerPOMDP) is ambiguous.

Candidates:
  convert_o(::Type{N}, v::AbstractArray{F}, problem::Union{MDP, POMDP}) where {N<:Number, F<:Number}
    @ POMDPs C:\Users\mrkos\.julia\packages\POMDPs\XBTe5\src\pomdp.jl:184
  convert_o(::Type{O}, o, problem::POMDP{<:Any, <:Any, O}) where O
    @ POMDPTools.CommonRLIntegration C:\Users\mrkos\.julia\packages\POMDPTools\XBP7W\src\CommonRLIntegration\to_env.jl:110

Possible fix, define
  convert_o(::Type{O}, ::AbstractArray{F}, ::POMDP{<:Any, <:Any, O}) where {O<:Number, F<:Number}

Stacktrace:
 [1] top-level scope
   @ c:\Users\mrkos\scth\projects\julia\psenisar_ukazka\dev\POMDPModels\test\tiger.jl:30

Maybe the suggested fix should be implemented in either POMDPs or POMDPTools?

I thought that qualifying the call as POMDPs.convert_o in the test could fix this, but that does not work.

[zsunberg]

Yeah, I'll try to fix it in POMDPs now.

[zsunberg]

Should be fixed now!