Your calculation of $F_X(x) = \Pr[X \le x]$ is wrong because $X \in [-1,2]$. Instead, $$\int_{t=\color{red}{-1}}^x f_X(t) \, dt = \int_{t=-1}^x \frac{t^2}{3} \, dt = \left[\frac{t^3}{9} \right]_{t=-1}^x = \frac{x^3 + 1}{9},$$ consequently $$F_X(x) = \begin{cases} 0, & x < -1 \\ \frac{x^3 + 1}{9}, & -1 \le x \le 2 \\ 1, & x > 2. \end{cases}$$
You calculation of $F_Y$ is also incorrect for the same reason; i.e. $X$ may be negative. However, it is not necessary to compute $f_Y$ or $F_Y$ explicitly. We simply compute $$\operatorname{E}[Y] = \operatorname{E}[X^2] = \int_{x=-1}^2 x^2 f_X(x) \, dx = \int_{x=-1}^2 \frac{x^4}{3} \, dx = \left[ \frac{x^5}{15} \right]_{x=-1}^2 = \frac{33}{15} = \frac{11}{5}.$$