Wedge product and cross product - any difference?
Solution 1:
I would have left this as a comment, but apparently I can't comment yet.
To answer your final question, why introduce the wedge, the point is that the wedge product (as explained in the Wikipedia article) is a notion that generalizes to $\Bbb R^n$ and indeed any vector space---in general the output is what is called a "bivector". Now, it just so happens that in $\Bbb R^3$ there is a natural identification between bivectors and vectors, and the vector that corresponds to the (bivector) wedge product is indeed the cross product, but this does not work in other vector spaces.
So, to sum up, the notion of cross product is specific to $\Bbb R^3$, whereas wedge product makes sense in any vector space.
Solution 2:
Yes, this definition is chosen for a reason, as the unique solution to a pedagogical problem.
Do Carmo's definition is awkward and redundant in 3 dimensions, but it is the only one among the usual definitions for the cross product that when generalized to $n$ dimensions (there is a cross-product of $n-1$ vectors in $R^n$) is rigorous, visibly basis independent, and produces a vector in the same space --- all without a long digression on multilinear algebra, duality, and other sophisticated theory. Consider the other definitions:
The right hand rule does work in $n$ dimensions, and is coordinate-independent, but it is hard to describe without a formula which hand is the "right" one, and verifying that the result is multilinear can be awkward if one does not have a reliable formalism of $n$-dimensional Euclidean geometry. I doubt most students would consider this approach rigorous even if, in principle, it can be made so.
The exterior algebra relies on theory not known by all audiences of Do Carmo's book (e.g, engineers). It does not produce a vector in $R^n$ without even more theory.
The definition as a determinant of the vector coordinates can be confusing, with vectors and scalars mixed together as entries of one matrix. It is also geometrically meaningful only when proven independent of the choice of (orthonormal) basis for the coordinates. In 3-dimensions this is known by the equivalence with the right hand rule and by drawing pictures, techniques that require a long development to perform in $n$ dimensions.