The following proof is a modification of some results from the paper The Jordan Curve Theorem Via the Brouwer Fixed Point Theorem by Ryuji Maehara (The American Mathematical Monthly vol. 91/10, 1984, pp. 641--643).
Note that by applying the transformation $t\mapsto 2t-1$ we may consider the interval $[-1,1]$ instead of $[0,1]$ and continuous functions $\omega,\tau\colon[-1,1]\to[-1,1]\times[-1,1]=:I$ such that $\omega(-1)=(-1,-1)$, $\omega(1)=(1,1)$, $\tau(-1)=(1,-1)$, $\tau(1)=(-1,1)$. If $\omega(s)\not=\tau(t)$ for all $s,t\in[-1,1]$ the continuous function $N$, $N(s,t) :=\Vert\omega(s)-\tau(t)\Vert$, where $\Vert(u,v)\Vert:=\max\{\vert u\vert,\vert v\vert\}$ is different from $0$ for all $s,t$. Thus $F\colon I\to I$, $F(s,t):=\frac{\tau(t)-\omega(s)}{N(s,t)}$ is well-defined and continuous. Then by Brouwer's fixed point theorem there would exist $(u,v)\in I$ such that $F(u,v)=(u,v)$.
Since $\Vert F(u,v)\Vert=1$ by definition of $F$ this implies $\vert u\vert=1$ or $\vert v\vert=1$. Let $\omega=(\omega_1,\omega_2)$, $\tau=(\tau_1,\tau_2)$ with continuous functions $\omega_i,\tau_j\colon[-1,1]\to[-1,1]$. We have four cases.
- $u=-1$. Then $\tau_1(u)=-1$ and $-1=\frac{1-\omega_1(v)}{N(-1,v)}$ implying $1-\omega_1(v)<0$ or $\omega_1(v)>1$, a contradiction.
- $u=1$. Then $\tau_1(u)=-1$ and therefore (similar as above)
$-1-\omega(v)>0$, again a contradiction.
- $v=-1$. Then $\omega_2(v)=-1$ and finally $\tau_2(u)+1<0$, again a
contradiction.
- $v=1$. Then $\omega_2(v)=1$ which implies the contadiction
$\tau_2(u)-1>0$.
Thus there are $s,t$ such that $N(s,t)=0$, as desired.