tl;dr: There is currently no 'foolproof/plug-n-play' solution to this problem (yet!). All of the currently available options have trade-offs which must be evaluated before jumping in.

---

This is a huge problem which is frequently discussed in the timelapse community. As of this writing there are not 'foolproof' solutions, though there are a good number of us who are attempting to create various 'plug-n-play' solutions to the problem.

As stated elsewhere, all automated modes tend to introduce a rather annoying level of flicker, especially during dawn and dusk when the light changes especially rapidly. While to some extent this can be managed with software that 'equalizes' the light levels across multiple frames and reduces flicker in post production for a short timelapse, the longer the timelapse is, the more problems are introduced in that the level of computer which can accomplish this task without simply breaking down and weeping openly is quite expensive. Additionally, whether these types of software are capable of doing a 'reasonable' job at the task of removing flicker is hotly debated by some.

One solution that a lot of us have had success with is to take bracketed frames instead of singles. This gives the option of using fading in post production to adjust in a relatively seamless way for the extreme changes that can occur. Depending on your camera you may be able to adjust for up to +4/-4 stops, giving you an effective dynamic range of 8 stops (full day to full night is approximately 12 stops, YMMV depending on time of year, location on the planet, etc.). Shoot that in RAW (gulp!) and you can batch process to add stops even beyond that. You'd have to experiment in order to see if the motion of your vehicle causes too many syncing problems, but I suspect that since you'd be fading between large chunks of frames, the syncing problems wouldn't be that bad. Obviously this is less than ideal in other ways, namely file size, so as always there is a trade-off to be evaluated.

I believe it would be possible to code a software solution to provide intervalometer functionality as you asked about. The problem that would need to be solved is that the software-based intervalometer would also have some sort of powered light sensor which would tell the computer what to adjust the shutter speed to. Either that or you could (hypothetically) build an algorithm that generates an appropriate curve to simulate the falloff of light as full-day goes to full-night. Then this algorithm could be used to automatically adjust shutter speed in a 'dumb' manner (e.g. it doesn't actually know what the light level is). All of this leads to the 'nuclear option' of options (at least as of this writing, anyway)...

What I have chosen to do is probably even an order of magnitude more extreme than anything above... I built my own intervalometer which includes a built in light meter and can adjust the shutter speed shot to shot as the lighting levels change. This is- by far- the most reliable way I've found to handle changing lighting conditions, and with it I've been able to get flicker-free full-day to full-night timelapses. But naturally the trade-off is that this is a home-brew device, so an electronics background (or a willingness to learn), the ability to code, or re-purpose others codes, a couple-hundred dollars worth of electronics parts and a soldering gun is required for this solution.