Wandering Robots in the Wild
spectrum.ieee.org
spectrum.ieee.org
As a kid, a slightly distressing part of Star Wars was R2D2 getting zapped by a band of cute but spooky-eyed scavengers who kidnap and sell robots for spare parts!
Robots may survive in well lit, surveilled streets of Earth's cities, but they won't need to wander as far as outer-rim desert world of Tatooine to get into trouble.
A situation much like the early days of network hacking, but in the real physical world is latent. Despite benevolent anthropomorphism, robots aren't much loved by humans. They're packed with valuable electronics. But much more fun... they can be captured, hacked to perform the bidding of a new master and released back into circulation. Iran capturing the RQ-170 stealth drone in 2011 was almost inevitable but portends a new frontier in digital defence in physical space. Ultimately this leads to robots that can "defend themselves", and so to the obvious escalation of force between humans and machines.
For this reason I predict that, like domesticated dogs, robots have limited room to rove, close to protective masters.
That brings to mind the hitchBOT [1] experiment from 2013 to 2015. It did well - helped by media attention - in parts of Europe and Canada, but later came to grief in Philadelphia [2].
[1] https://en.wikipedia.org/wiki/HitchBOT
[2] HN comment thread on HitchBOT destroyed in Philadelphia, ending U.S. tour
After all real dogs and cats usually aren't in danger of kidnapping even though you could sell them or eat them or something. Indeed if they're anything like real animals, the biggest danger will be inattentive drivers.
* Have you considered not flying your military drones over other countries?
I suspect the bigger danger will be vandalism. I will admit that the robots that wander my local grocery store looking for messes to clean up (or at least to alert humans to) make me irrationally angry. They get in my way and distract my attention, and it makes me want to just kick the *$%# out of them. I suspect I'm not the only one who feels this way, and the more robots get 'out in the environment' without anybody watching out for them, the more people will damage them on purpose.
I do feel a sort of spirit in the machine for these new AI gadgets, even though I am aware of the reality of it.
Brazil has some experience with people being nasty to machines.
1. Robotic bank tellers (atm): we had a huge crime wave where people would literally blow them up to steal contents. It got so violent that some places refused to have one nearby, some explosions demolished buildings, and gun battles sometimes happened, one of them even involved people using anti-material sniper rifles and left the surroundings completely demolished and several corpses destroyed in gory ways (like people cut in half by bullets)
2. Only places where you can play arcade games are very secure (like airports), I asked a manufacturer about it, and he said any arcade game in a bar for example invariably get destroyed as people steal its components (chips, cables, etc... people steal the valuable silicon parts for sale, and the metallic ones like cables and connectors to melt down)
The method of choice in my part of the world is to use a JCB (or other brand of digger). Just a cursory search will show examples such as [1] and [2] in England and [3] in Ireland.
From [2] (with my apologies for it being the Daily Mail): There have also been a spate of similar robberies elsewhere in the country.
In November thieves used a digger to smash through the brick wall of Warner's Budgens in Moreton-in-Marsh, Gloucestershire, before fleeing the scene with the ATM.
In the same month, thieves pulled away bricks leaving a gaping hole in the side of a Co-op supermarket in Sapcote, Leicestershire, which took just 20 seconds.
Similar to the robbery in York, the digger was left at the scene.
Thieves also ploughed a stolen JCB digger in to a Marks & Spencer store before stealing the cash machine at Malvern Retail Park in Worcestershire.
[1] (England) https://www.independent.co.uk/news/uk/jcb-thieves-ram-raid-b...
[2] (England) https://www.dailymail.co.uk/news/article-2921779/Now-s-hole-...
[3] (Ireland) https://www.rte.ie/news/courts/2022/0304/1284407-atm/
Maybe, but robots will have something that domesticated dogs, and their wild counterparts, don't have - the ability to communicate and effectively (perfectly?) share information. Just like a driver would ask a gas station attendant for directions, a lost robot could ask a local one for information on its surroundings and use that to plot a course. My Roomba may not need to leave my house, but it could tell a passing delivery bot when it has arrived at what I consider my front door.
Why people fuel this fantasy of roaming robots being stolen?
Of all valuable things that can be stolen, old cars are the easiest choice for a thief and robots are the worst.
An AI device filled to the brim with sensors, cameras, GPS and with remote connectivity is the perfect go-to-prison card.
> Ultimately this leads to robots that can "defend themselves", and so to the obvious escalation of force between humans and machines.
Is this scifi worldbuilding or reality? No, your new car is not going to incinerate thieves like in Robocop 2.
Is therefore filled to the brim with valuables that an old car doesn't have.
Take for example the ZiGGY [1]. I predict that within 24 hours of it being in service, someone will lift one of them into a van, drive away with it to an abandoned shed, and tear it down for the batteries inside, before the owner can do anything about it.
[1] https://www.businesswire.com/news/home/20220614005357/en/EV-...
>Is therefore filled to the brim with valuables that an old car doesn't have.
Because cameras and GPSes are expensive now? Is a phone more expensive than car from 20 years ago then?
- microswitches that trigger if it ran into something
- a set of photodiodes for detecting "cliffs", i.e. keep it from falling down the stairs
- another photodiode for sensing the dock's IR beacon
- something that looked like a step counter (but it kept getting gunked up, yet never impacted the robot's behaviour - which makes me believe it wasn't actually used for anything)
No cameras, LIDARs or internet connection, and I'm pretty sure no local mapmaking either.
Movement was essentially random, with a small set of heuristics and preprogrammed patterns. When you were actually watching the robot, the random movement was kind of ennerving - it would frequently clean the same spot over and over again, while completely missing another spot right next to it.
However, when setting the robot on a daily cleaning schedule, the random movements would even out and the room would be cleaned pretty well in the end.
How? By noticing one day that the black areas of my striped carpet were inexplicably accumulating more dirt than the rest of it. Took me some time to find out that the robot treated black floor as "hole in the ground" and made an active effort not to drive over it...
> Mine also regularly blew through the IR "walls" and got stuck under a dresser, waiting for me to return home with it's familiar "help" beeps.
Yup, got those often enough as well...
source: For my masters degree I wrote a paper on the grounding problem (linking words said to physical objects), specifically the spatial reasoning component ("please give me the book on the table in the corner). We were able to get very simple commands to work, but anything remotely complicated, especially with a temporal aspect or things that can't immediately be seen (keep going until you see the red building, then turn left) is the realm of future research
Relative size, position, etc can be tricky to determine as they aren't just classification problems but require segmentation and spatial reasoning. Especially taking into account any difference in perspective between the user and the robot, as well as context (what does "above" mean on a map vs in a tower of blocks, or what does "left" mean if you are facing me)
Starship Robotics has a pilot in Milton Keynes in the UK for their delivery robots, and came across the problem of pedestrian crossings where you need to push a button to change the traffic lights & cross. Rather than engineering a solution that required changing the crossings or putting some complex robot-arm on to push the button, they just play a pre-recorded message begging humans to press the button for it.
It's easy to devise complex solutions which you can't implement - but it sometimes involves a bit more creativity when you constrain yourself to simple and implementable solutions!
https://arxiv.org/abs/2207.04429 https://arxiv.org/abs/2204.01691#google https://arxiv.org/abs/2207.05608#google https://arxiv.org/abs/2207.05608#google https://arxiv.org/abs/2109.01115 etc
This is a fairly accurate description of a "guide" i got from a nice guy after asking him where my hotel may be... and now imagine the software that must be involved to decode THIS...