There are many reasons why this isn't ever happening. A very big one is that such a 'battery' would be producing heat all the time. Say your device has 10W peak demand, and your radioisotope thermal generator (nuclear battery) has efficiency 10% (better than we've yet achieved), then you'd need an RTG which was emitting 100W of heat all the time. (On the plus side, it would do a fine job of heating the interior of your car on cold days.) (If your device only uses 10W occasionally, you could pair a 1W output RTG with rechargable batteries, but now all you're saving yourself is the need to plug it in each night.)
* Cost - even with efficiency of scale, producing radio isotopes will be very expensive
* Scaling - the technology works (sort of) for 100W power generation, it may be hard to scale down to 10W or 1W
* SIze - a 100W RTG is the size of a person.
* Safety - they contain really nasty radioactive sources. If you use alpha emitters, you can make them 'safe' with very thin shielding, but once the material escapes into the environment (e.g. in a house fire, or someone chops the battery with an axe) it is very nasty indeed.
Yes, future technology can help somewhat with any of these - but it needs to improve all of these problems, each by many orders of magnitude, before nuclear batteries will be practical.