<: POMDP (e.g. QuickPOMDP)
transition(m, s, a)
observation(m, s, a, sp)
reward(m, s, a, sp, o)
@gen(:sp, :o, :r)(m, s, a)

<: Updater
update(updater, b, a, o)

initialize_belief(updater, b0)

<: Policy
action(policy, b)

<: Solver
solve(solver, m)

<: Simulator
simulate(sim, m, policy, updater)