My engagement with this topic was piqued by this question, in which the OP MathMan was seeking help in applying the principle that two plane curves with identical curvature function (I will make this more precise in what follows) are themselves identical "except for probably their position in $\Bbb R^2$" (sic). As I attempted to answer MathMan's concerns, I became more and more aware that the underlying concept was worthy of address in and of itself. Specifically, I began to wonder just how it might be proved. In particular, I wanted, and still want, an analysis/proof of the assertion that two curves the curvature functions of which are the same are "congruent" in the sense that one may be made pointwise identical to the other by a rigid motion of $\Bbb R^2$. In formulating a precise statement of this result, I searched math.stackexchange.com in the usual manner for related questions, but found nothing which seemed exactly on point, so I am proceeding to ask it here.
Having said these things, I turn to my
Question: Let
$I \subseteq \Bbb R \tag 1$
be an open interval, not necesarily bounded, and let
$\alpha, \beta: I \to \Bbb R^2 \tag 2$
be regular, arc-length parametrized curves with curvatures
$\kappa_\alpha, \kappa_\beta: I \to \Bbb R^+ = \{r \in \Bbb R, \; r > 0 \}, \tag 3$
as defined in the Frenet-Serret equations,
$\dot T_\alpha(s) = \kappa_\alpha(s) N_\alpha(s), \; \dot T_\beta(s) = \kappa_\beta(s) N_\beta(s), \tag 4$
where $N_\alpha(s)$ and $N_\beta(s)$ are the unit normal fields to $\alpha(s)$ and $\beta(s)$, respectively. Then if
$\kappa_\alpha(s) = \kappa_\beta(s), \; \forall s \in I, \tag 5$
it follows that there is an orthogonal transformation $O$ of $\Bbb R^2$ and a vector
$\vec v \in \Bbb R^2 \tag 6$
such that
$\alpha(s) = O\beta(s) + \vec v, \; \forall s \in I. \tag 7$
Throughout, by a curve I will refer to a $C^2$ map $I\to\mathbb{R}^2$ with nonvanishing first and second derivatives, where $I\subseteq\mathbb{R}$ is an open interval. For these curves, the unit tangent and normal $T,N$ are always well defined and continuous. Also, I'll use $$ R_\theta:=\begin{bmatrix}\cos(\theta)&-\sin(\theta)\\\sin(\theta)&\cos(\theta)\end{bmatrix} $$ As a shorthand for rotation matrices
Since $T$ and $N$ are nonvanishing, orthogonal, and continuous, it must be the case that $N=R_{\pm\pi/2}T$. We may define the orientation of a curve as counterclockwise if $N=R_{\pi/2}T$ and clockwise if $N=R_{-\pi/2}T$.
Your claim can be proven by showing the resulting ODE has unique solutions, and then using a bit of plane geometry:
Lemma: Fix a strictly positive and continuous function $\kappa:I\to\mathbb{R}^2$, and fix $t_0\in I$, $\gamma_0,\dot{\gamma_0}\in\mathbb{R}^2$, with $\|\dot{\gamma}_0\|=1$. There is a unique counterclockwise (alternately, clockwise) unit speed curve $\gamma:I\to\mathbb{R}^2$ with curvature $\kappa$ which satisfies $\gamma(t_0)=\gamma_0$ and $\dot{\gamma}(t_0)=\dot{\gamma}_0$.
Proof: The condition that $\gamma$ has curvature $\kappa$, along with the fact that $N=R_{\pi/2}\dot{\gamma}$ (since $\gamma$ is unit speed and counterclockwise) allow us to write the conditions as a linear second order initial value problem: $$ \ddot{\gamma}(t)=\kappa(t)R_{\pi/2}\dot{\gamma}(t),\ \ \ \ \ \gamma(t_0)=\gamma_0,\ \ \ \ \ \dot{\gamma}(t_0)=\dot{\gamma}_0 $$ This IVP has a global solution, given by $$ \gamma(t)=\gamma_0+\int_{t_0}^tR_{\theta(\tau)}\dot{\gamma}_0d\tau,\ \ \ \ \ \theta(t)=\int_{t_0}^t\kappa(\tau)d\tau $$ And since the differential equation is locally Lipschitz, this solution is unique. To see this, let $\gamma,\lambda$ be two solutions. The set $S=\{t\in I:(\gamma(t),\dot{\gamma}(t))=(\lambda(t),\dot{\lambda}(t))\}$ is open by the Picard-Lindelöf theorem, but its complement $I\setminus S$ is also open, since $(\gamma,\dot{\gamma})$ and $(\lambda,\dot{\lambda})$ are continuous. Therefore one of these sets must be empty, and since $t_0\in S$ by initial conditions we have $\gamma=\lambda$. The clockwise case is true by exactly the same argument with all of the rotation matrices inverted. $\square$
Completing the proof requires a few more straightforward facts about curves in $\mathbb{R}^2$; namely
Given these facts, and two curves $\gamma,\lambda:I\to\mathbb{R}^2$ with equal curvatures, we may choose a base point $t_0\in I$ and choose a rigid motion $g:\mathbb{R}^2\to\mathbb{R}^2$ which takes $(\gamma(t_0),\dot{\gamma}(t_0))$ to $(\lambda(t_0),\dot{\lambda}(t_0))$, chosen to be orientation preserving if $\lambda$ and $\gamma$ have the same orientation, and reversing otherwise. By the previous lemma, $g\circ\gamma=\lambda$.