CAViaR modelling

I am wondering how to estimate in R CAViaR modeldescribed by R. Engle
and S. Manganelli in their article avilable at R. Engle's webpage:

http://pages.stern.nyu.edu/~rengle/

Am I correct, that i should start with writing function describing the
process (adaptive or indirect garch), then call:


quntile <- quantile(y[1:n], probs=0.05)

library(quantreg)
nlrq(y ~ indirect_garch(y), quantile, tau=0.05, 
             control, trace=FALSE,method="L-BFGS-B")

where n is a 30% of the sample, to compute the unconditional quantile
for start and the indirect_garch is a function described in the
article. Is the objective function in nlrq the same as in the article?
I would appreciate help on this field a lot. I think that CAViaR
models can give very good results in estimating Value at Risk for a
single risk factor.

Best regards,
Wojtek