To characterize pulse width in the femtosecond-picosecond range you need an autocorrelator. Ape systems have some nice ones and easy-to-use

So this really depends on what you want. If your just want the basic information an auto-correlator will do the job, you can build your own, just need a crystal some optics and a translation stage. (simple! :-)) you can do this for an order of magnitude less money than paying Swamp Optics $10k. However, the auto-correlator will only give you symmetric information, i.e. you won't know if that feature is a prepulse or a postpulse. But, for that matter you can build your own FROG for the same $1000, but then you need software to analyze the spectral info, which might be great for a student to do, but isn't really thesis work. In any case, for more detailed info you need a cross-correlator. (Or a FROG or SPIDER). This much more involved, and is commercially available. Del Mar Photoics makes a good cross-correlator, their web site SUCKS, but their products are good. But if you need spectral and phase info, or worry about pulse stability buy the FROG. Del Mar makes one too, have no idea how it compares to the GRENOUILLE. Hope this helps.

http://www.dmphotonics.com/

AC is the most common technique despite the fact its the least informative about the phase, especially high order dispersions. FROG (or even better XFROG) and SPIDER are two common ways that are more practical in retrieving phase information. However, I use something called MIIPS (multiphoton intrapulse interference phase scan) which is a pulse characterization and correction method. I have used all the methods (other than SPIDER) and I can definitely tell that MIIPS is the most practical method in terms of retrieving pulse characteristics and phase information. About Grenouille, I haven't used this method and can't comment about it.

Optical sampling oscilloscope is a simple solution. For instance, AQ7750 (yokogawa) cost you about 28k USD from ebay

I'm a current user of GRENOUILLE 8-20.

The characterization method absolutely depends on the pulse duration you want to measure.

FROG or GRENOUILLE works great for around 30-fs pulses from my experience. (I used to use home-made single-shot FROG system too.) If you have enough budget and the specifications of GRENOUILLE fit your system, I definitely recommend to purchase one. It is very easy to align. Also, Swamp optics provides excellent software that shows you real-time retrieval information of pulses you are measuring. It has reduced a lot of work characterizing the laser pulses.

If you need just pulse duration information or the pulse duration is longer than 100 fs, simple autocorrelator would be a good solution because it does not have very complicated structure. On the other hand, if it is less than 10 fs, even FROG is not able to characterize the pulse correctly. FC spider is known to measure down to 5 fs. However, you will need other techniques using attosecond pulses such as FROG-CRAB or pettahertz optical oscilloscope (I'm using this for 5-fs) if it is less than 5 fs.

Although I didn't test experimentally (I just did simulation work), dispersion-scan method seems to be also simple and accurate. It has extremely simple setup (so it is very cheat to build) but you need to code a software to retrieve spectral phase.

Your question has no clear-cut answer if you don't provide additional information.

If it fits your situation, from my own experimental experience, I can really vouch for dispersion-scan types of measurements (chirpscan, d-scan, MIIPS). They can be really accurate. In my opinion, the best thing about them is that they are quite difficult to mess up badly, and the d-scan is probably the cheapest pulse measurement ever, however, you either need to code the retrieval software yourself (I did that, it's not particularly easy, but perfectly possible) or buy it from, for example, Sphere Photonics.

I think it should also be noted that in case you have pulses of different complexity and duration in your setup, it may be useful to try some cross-correlation (possibly spectrally resolved) techniques!