Autonomous robot to stave off spotted lanternflies | Hacker News Reader

Autonomous robot to stave off spotted lanternflies

cmu.edu

35 pointsgeox21 comments

Fun student project, but let's say you do spend millions of dollars on R&D to get this into production and released into the wild to target lantern flies eggs.

* What percentage of eggs in a given area is it expected find and destroy? If it's anything less than ~100% then you are guaranteeing that within a few generations (years) you will be dealing with a population of lantern flies naturally selected for traits that thwart this robot predators hunting strategies.

* Taking the cost of R&D, production, deployment, and operation of the first generation, how much economic damage was mitigated before this generation of robot becomes ineffective and needs to be "evolved" to deal with this newer generation of lantern flies?

Imo, human beings can't compete in the game of Life with these kinds of strategies and we'd be better off selectively breeding and/or genetically engineering existing predators to preferentially target lantern flies and their egg sacks, instead of building fully autonomous robot predators from scratch.

Lanternflies seem to have died out from relevance in PA, at least compared to the peak a few years ago where they were everywhere. It was particularly shocking to see them clogging window gutters of builds 15 stories high.

This is my experience as well locally in SE PA. 2018-19 seemed like the peak and they were literally everywhere, but I’ve only seen a handful of adolescent lanternflies this year.

Interesting robot. But would it be efficient in actually reducing the amount of flies in a significant way? I am thinking that trees can be high and it seems not being able to reach much higher than a human would. Perhaps some kind of climbing robot could work here? Or a drone on a leach?

Why are these types of robots always so slow? Are they waiting for feedback on position before moving again?

They're waiting for genetically manipulated true breeding male flies.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9274970/

Fast movement = higher energy collisions. Robot arms can move very fast, but usually you only optimize for that in production. At that point the system should be collision-free.

Which API did they use for location extraction?

I am pretty sure they use no API at all, all the processing is local.

The guy in the video says "We use OpenAI Step API to extract the locations of the detections".

I am unaware of that API and could not find it.

Looks very impressive, but the robot they built looks expensive to me. I would hazard a guess that it wouldn't be much more cost effective than hiring a human to do this job.

I suspect that as societies all over the world age further, you'll be hard pressed to find anyone willing to do this job.

I had a plumber replace some pipes in my apartment the other day and he said that he agreed to take this job despite being retired, because there's simply no one available to do it.

His effective hourly rate was around 50% more than I make.

Your particular pipe job? Or the job as a plumber?

My particular pipe job. He was done in a little less than an hour and all that including cutting through the plasterboard housing the thing.

I absolutely believe this. I’ve had wait times quoted in weeks for plumbing work in the US Midwest, with electrical work lead time quoted in months. The trades worker participation rate decline is real.

My friend poured the concrete base for his heat pump by himself because apparently that's not part of the service when installing one and finding a guy to do it was nigh impossible.

At the moment I dread doing any house maintenance - especially the sort for which you need certification.

As a property owner, my recommendation (time permitting), is to become a scholar of the building sciences. Assume that any work short of major renovations or structural could be your problem.

The mentioned friend says that he holds multiple PhDs now that he's done with the main part(phrasing intentional).

The sauna didn't go exactly as planned though. Straight lines are hard.

In most instances you have to remember you’re paying not only their salary but their employment taxes, benefits, tools depreciation, transportation depreciation, advertising, and admin. Plus the time it takes to travel, etc. One guy I know told me he has to charge $250 base to show up (then hourly once work starts) otherwise he can’t make money.

But yes, there is a huge shortage and it’s growing. Where I live tradesmen who work for a small shop pull 6 figures, get a matching 401k, and free healthcare. They get poached left and right and if you’re trying to run a business you can’t have that so you pay.

Honestly it sounds pretty much like the tech industry right now.

Side note: politicians and ivy educated journalists want many things electrified by 2035. Even if every electrician worked 60 hour weeks and took no vacation we still couldn’t hit those goals. Not even close.

My two year old grandson is going to be fed on a steady diet of information about the virtues of becoming a plumber as his career over the next 16 years.

To me, plumbing is just low tech enough to stay automation and AI free for the long haul. Also I don’t ever foresee a time where people will not want clean water brought in to their homes and have their shit removed.

Plumbers are going to own the world in 20 years.

It's a prototype. The first few iterations of everything are expensive. Then mass production steps in.

Reply on news.ycombinator.com