A qualified guess is that you should use a floating point vector here, i.e. u0 = [155.0, 107.0, 93.0]. I think the output vector is allocated based on this input, i.e. a Vector{Int}, and therefor you get this error when you try to insert something that is not convertible to an Int, i.e.
julia> A = [1, 2]; A[1] = 1//2
ERROR: InexactError: Int64(Int64, 1//2)