This is my reasoning (for convenience, the particle now starts at 0.5 and has a speed of 0.5):
Let's start with a particle that, at the start, decides to either go up or down. In this case, After one second it'll have a 50% chance of being at 0 and a 50% of being at 1.
If, instead, the particle first decide at 0 what direction to have, and then decides again after 0.5s, after 1s it'll have a 25% chance of being at 0, a 50% chance of being at 0.5, and a 25% chance of being at 1.
In general, if the particle decides every $1/n$ s what direction to have, it'll have a probability of landing on $a/n$ of $\frac{\binom{n}{a}}{2^n}$ (for $0\le a \le 1)$. So, let's define $g$ as $g(\frac{a}{n})=\frac{\binom{n}{a}}{2^n}$, so $g(a)=\frac{\binom{n}{an}}{2^n}$.
The problem with $g$ is that $\int_0^1g(x)dx\ne1$; to fix this problem I define a new function $f$ as $f(x)=\frac{g(x)}{\int_0^1g(x)dx}$ (now, $\int_0^1f(x)dx=1$); perhaps it's not a very rigorous step, but I feel like it makes sense.
The original problem is just the special case $\lim_{n\to \infty}$; graphing $f(x)$, it's pretty evident that the function approaches the dirac delta function (here's some graphs:https://www.desmos.com/calculator/nwew2tplqj).
(Does all of this make any sense?)