Ive thought about it...you need to be able to control the speed of the fish feeder and that will require alteration of the product. You could, probably, just drop a pellet every 30 seconds with less effort than altering the machine to your needs and testing it to work but you would have to monitor the boil continuously (obviously).
If you could alter it to adjust speed, then pulverize your hop pellets to homogenize your hops, then use your uber math skillz to create an algorithm to predict your IBU's you might have a badass auto-hopper. But then you lose the ability to accurately estimate your IBUs and the ability to use different hops at different times whilst maintaining repeatability between batches. If those issues dont bother you, then you should try this thing out!
One of the main problems I came up with is that the water vapor from your boil will mix with your hops and provide you with a sticky-hop-pellet scenario that is hard to avoid and even harder to deal with once its occured.
Caligione used the "vibrating football field" method to achieve what you want without the precision or the sticky-hop-mess.
I loosely began designing a fish-feeder based hopper but the water vapor situation kept leading me to dead ends. I chalked it up to a "great idea with high difficulty and low return" project.
I hope you have better luck than I did.