Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Headless, dog-sized robot to patrol Alaska airport to prevent bird strikes (sky.com)
50 points by Brajeshwar on March 30, 2024 | hide | past | favorite | 47 comments


Seems like a pretty good use case for an automated robot like this. It's nice to see a practical use-case for these advanced autonomous robots that doesn't involve war.

So as I see it, this is like a fancy all-terrain Roomba. They dress it up like a Fox or Coyote and it wanders around and patrols the airport grounds, scaring away birds and generally preventing them from nesting in the area. Birds aren't going to nest in areas with active predators. So if this looks like an active predator making constant patrols, then birds will just go somewhere else to roost.

Ultimately it saves the birds from being killed by planes or euthanized by other means. They are not harmed with this patrolling robo-fox, it just adds intimidation factor.

Presumably, the robot will also be smarter than a Roomba and would know to avoid running onto the runway or other critical areas.


It’s not autonomous and it is not smart, you are giving a lot of credit for an expensive toy.


Betcha it won’t take long for the birds to realize the robot is no threat.


Depends on the bird. I once tried a remote doorbell on a bird table to scare off large birds (pigeons); by about the third press they learnt to ignore it.

In this case could be geese, which seem more cautious (then again if it's corvids, they'll probably figure out how to reprogram the robot).


Depends on the season, if it's spring the geese will just kill the robot.


It doesn't take very much to permanently scare a bird away from a nest they're in the process of building. They're much more sensitive to threats when establishing a nest.


Depends on the species. Ever met pigeons?


I was actually thinking of pigeons/doves when I wrote that. Every year I get several nests around my property and if I get too close before there are chicks, the pigeons abandon the nest.

I have to carefully track them to figure out which nests belong to mockingbirds and which ones belong to pigeons so I can chase the former away and let the latter do their thing.


Mine were happily doing their thing even while I started the same balcony. Eventually I started sterilizing their eggs by cooking and placing them back (they do leave for a few minutes when actively shooing) and after a few failed clutches they are noticeably less eager to try again in that place.


Have you tried scaring them away before they lay the eggs or is it the same result? I never noticed eggs in the nests that were abandoned (I would put them in an incubator if I did)

The term "pigeon" refers to the whole Columbidae family with over 300 birds so YMMV. Mine are specifically mourning doves which are different from the pigeons I associate with cities (feral domesticated rock doves)


In France pigeons are protected and you have to let them nest (more precisely you can't move the nest once it's there and you cannot harm the eggs)


They, and the deer, will tend to be in the runway area to avoid the “dog”, which will stay off the runways.


I imagine that for any small-to-medium sized avian a ~100lb robot could very easily become a threat.


okay but why not, just, use dogs or coyotes

not joking. Dogs can be trained. Coyotes… just set up some feeding pens full of carrion at either side of the airport, and set up some doors that open at intervals (there, scratched that robot itch).


https://m.youtube.com/watch?v=-QxZv_wJ1aA

Airport bird dogs are already a thing. Seems like a dream job for a dog.

But you need a full-time handler, a car, a kennel, training and eventually a replacement. They can’t work 24/7, and if they are out sick for a day the birds are back. Essentially the argument for any kind of automation.

This comment is like looking at a scarecrow, and saying “why not just go and stand out in your field all day?”

And feeding wild coyotes unlimited food, near people, comes with its own set of problems.


It is very expensive to train a dog and maintain its training. Like $50k starting price and $20k every year after.

Dogs are incredibly stupid, they like to step in front of moving vehicles. Very bad for airport runways...


Perhaps because you don't want planes hitting coyotes on the runway.

You especially don't want to be the person responsible for a dog-strike. Don't mess with the FAA nor the SPCA nor the unions.


the noise is way too harsh for dogs at many times of operation?


I mean you can treat the coyotes like dogs on meth, considering they do exist as pets. That is, i thought just what you wrote but with no need to have extra carrion.


This appears to be a stock Spot with a paint job and some mission specific code?

https://bostondynamics.com/products/spot/


This Anchorage Daily article confirms that it's a Spot robot. The first image caption mentions Boston Dynamics. They just named this individual robot "Aurora".

https://www.adn.com/alaska-news/2024/03/27/alaskas-new-robot...


Back in my day, they'd just call the Pinkertons


Do scarecrows not work anymore? I find it difficult to believe that this is the most cost efficient-way to address the issue. I thought it was at least going to target the birds with high-powered speakers calibrated towards the birds hearing (which again might be easier to deploy in turret form), but this just seems like a joke or a parody of "tech investment" or something.

Granted, this does appear to be the test phase, but I'd hope there be some evidence this might work before they spend money on it.


Anyone who has a garden knows scarecrows don't work (more than a couple of days anyway).

Birds are surprisingly smart and they lean very quickly.


Did scarecrows ever work?


I'm just thinking aloud here. Scarecrows might make the pilots freak out too. This thing has to not look like something that doesn't belong near a runway.


I agree, seems like overkill especially if you need a person to manage the "dogs". Instead I'd create a track around the airport and have a "dog" like train that just travels along the track autonomously. You could randomize pattern and dog movements so the birds don't get suspicious.


This seems far from the best way to accomplish the goal, and has been turned into a big photo op, which to me comes across as a probable discount for publicity trade, indicating Boston Dynamics being surprisingly desperate for real world use cases. For all the progress in autonomous robots, maybe they’re actually in a sort of uncanny valley of generalization: more general than previously possible, but not general enough to unlock most tasks that previously couldn’t be automated.


What is with these weirdly conspiratorial comments?

A company sells their product for money to a government agency, and somehow this makes them "surprisingly desperate". Dude, they're not having the robot strip on the weekends as a side gig.


> Dude, they're not having the robot strip on the weekends as a side gig.

Beautiful.


If you really think not taking product news at face value is conspiratorial, you’re far too trusting. The transactions I’m theorizing here are very common and not illegal.

“Surprisingly desperate” seems to not be coming across as I intended, I meant only that it’s surprising that a product that should sell itself is instead searching for use cases, as weakly evidenced by this article but more strongly evidenced by their many past pivots without significant adoption.


What are more cost-effective ways of accomplishing the same goal?


Real dogs.



The photo from the article shows a pilot with controls (or technician with a monitor?). How autonomous is the robot?


Because it is not autonomous, it’s RC controlled. As long as you don’t see special sensors like lidars or cams, assume it’s not, and glad the article showed the guy too for the transparency.


I see at least 11 sensors in the top two photos of the article including 8 or 9 sensors on the front face of the robot and three sensors on the rear side mid body forward of the hind leg, which I assume is duplicated on the other side, so maybe 15 external sensors at least.

I also did photo journalism in college and these photos are from an AP News photographer. One of the things I learned in journalism class is to always try to get a person in your photo, as it helps humanize the story. So it may be an autonomous capable robot they showed in R/C mode for journalistic interest.


These sensors aren’t used for autonomy, this one for example you can tell it is capable however https://www.cbsnews.com/news/robot-dogs-to-patrol-southern-b.... OP pic sensors are fpv cams and collision avoidance, so of you drive it to a wall, it will stop before hitting it.

> photos are from an AP News photographer. One of the things I learned in journalism class is to always try to get a person in your photo, as it helps humanize the story

Interesting, I didn’t know that, thanks for sharing it!


Purely as a piece of photography that second photo is very cool with the sharp focus and the snow in the air.


Ah yeah that one is nice! I did photo journalism in college and always appreciate good journalism photos.


I had visions of robot dogs breaking down picket lines.


Why not use real dogs? They are really good at this.


Instead of a dog-sized robot, why not use a... dog?


How else they are going to milk millions of funds?! Throw in some buzz words and cool looking robots, welcome to the robotics industry!


You could definitely train a dog to do this and it would probably have a blast doing it all day.

Not enough dog investors pushing dog-tech hype though. It is a shame though, have you ever read about what their noses can do??


There are few robot welfare advocates complaining about robots doing what dogs love to do, though. Nobody is going to say robo-fido is being exposed to carcinogens or loud noises while it (merrily?) chases off birds.


Many airports in the states already employ full time dogs to clear birds.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: