I've stalled out on coming up with ideas to prove this.
Thought One:
Using $hb = aba^{-1}$ and $a^{-1}h = ba^{-1}b^{-1}$ somehow. However I can't assume $hb \in H$, I think.
Thought Two:
Using $bh = baba^{-1}b^{-1}$ and $ha^{- 1}= aba^{-1}b^{-1}a^{-1}$ somehow. But I can't assume those are in $H$ either.
I can't think of a plan to start solving this. Hints are welcome.
Let $C$ be the commutator subgroup $G/C$ is commutative, Let $p:G\rightarrow G/C, h\in H, g\in G$, $p(ghg^{-1})=p(h)$ implies there exists $u\in C, ghg^{-1}=uh$, since $u\in H$ we deduce that $H$ is normal.
If you do not know the notion of group quotient, use this: $ghg^{-1}=h(h^{-1}ghg^{-1})$ since $h^{-1}ghg^{-1}\in C\subset H$, we deduce that $ghg^{-1}\in H$.