Direction of the radiation reaction force (vs. velocity) The radiation reaction force is proportional to $\dot{\mathbf{a}}$, and can be derived from the the Larmor power using the concept of the conservation of energy. Looks like the radiation reaction force is a force acting in the opposite direction of the charge velocity, so that the charge would lose its kinetic energy. However, I don't see any relation between the directions of $\dot{\mathbf{a}}$ and $\mathbf{v}$.
The radiation reaction force can also be interpreted as the "self force". Using this concept, I cannot see whether the self force makes the charge lose or gain energy.
So, does the radiation reaction force always make the charge lose its energy?
 A: The radiation reaction force on a charged body is only approximately given by the Lorentz-Abraham force. Lorentz calculated force on a rigid charged sphere due to itself and the expression has infinity of terms; the Lorentz-Abraham force is only one term in the series.
There is no general relation between $\dot{\mathbf a}$ and $\mathbf v$. The Lorentz-Abraham force thus does not necessarily point in a direction opposite to that of velocity. This means the Lorentz-Abraham force does not necessarily decrease speed of the charged body.
A: Roughly speaking, the radiation reaction will make the charge lose energy. Consider, for example, two charges in orbit. The charges radiate, which causes them to slowly inspiral towards one another over time until they collide. In this case their speeds actually increase, but they also get closer together and lose eccentricity, which more than makes up for this. A closely analogous process obtains for binary black holes.
Now, in the black hole case this is not necessarily true (although practically speaking it essentially always is). It is possible for the self-force to briefly add orbital energy to the system. In exchange, the masses of the holes will decrease.
Again, roughly speaking, the effect of the self-force upon a single particle will be more or less what you expect; to carry energy off. 
Why do I keep saying "roughly speaking"? Well, as you note, the radiation reaction does not necessarily slow the particle down. The basic problem here is with the inappropriate mixing of the idea of a ``fixed particle" with a field theory such as electromagnetism. 
By considering a point particle, we have, in essence, demanded that part of the field (the field of the ``point") remain static. This is usually fine. But when making the self-force calculation we are essentially asking how that field ought to interact with itself. Since the field is by construction singular, we run into problems.
Now suppose we consider a non-singular blob of charge, and, importantly, we allow that charge to have a non-fixed rest mass. The self-force will always lower the energy, but there are multiple ways it might do this:


*

*by reducing the speed of the blob;

*by reducing the rest-mass of the blob;

*by dispersing or compressing the blob.


Since not all of these effects are possible in the point-particle approximation, we get inconsistencies. Many groups are attempting to address this, since the effect is important to black hole binary inspiral. See, for example, http://www.math.utk.edu/~fernando/barrett/bwald1.pdf.
