5 ms·

I love the idea of this because I hate lawn maintenance. But the idea of letting software I wrote run around with whirling blades of death underneath of it woul

I love the idea of this because I hate lawn maintenance. But the idea of letting software I wrote run around with whirling blades of death underneath of it would keep me from doing it.

vl 4y ago
They already make robotic lawn mowers. You can go and buy one right now.
Nextgrid 4y ago
A solution could be to just secure the area well enough to guarantee that 1) an uncontrolled machine can't escape and 2) people don't enter the area. Then you don't particularly care if the machine is "unsafe" if there's no way it can harm anyone. That's the approach most industrial robots take - instead of trying to deal with humans and not hurt them, just make sure there are no humans within reach of the robots.
- stevehawk 4y ago
  > 2) people don't enter the area That's harder than it should be. No amount of asking my neighbors has kept their kids out of my fenced lawn to retrieve their soccer balls. Which is actually an issue because one of my dogs is not particularly friendly. I'd gladly kick the ball back over if they just knocked on the door. We're not a neighbor-hate-neighbor situation, we actually get along quite well, but kids will be kids.
  - Nextgrid 4y ago
    Asking isn't enough, but proper fences should both make it impossible to get in without significant effort and also do a good job at containing the machine should it try to run away. It's up to you to run the numbers whether the cost of securing the area is less than what you spend (in terms of time/inconvenience/etc) manually mowing the lawn.
    - scarby2 4y ago
      Coming from the UK I'm always amazed that many (suburban) Americans don't have 6-8 ft fences surrounding their back yard.
      - Nextgrid 4y ago
        On the other hand, if everyone's yard is accessible, it's unlikely that yours will be the one subject to anything bad so I guess it's a number's game.
      - stevehawk 4y ago
        most of us outside of major cities or suburbia would lose our views.
  - ericd 4y ago
    An 8 foot mesh deer fence should do the trick ;-) it’ll also rebound lots of those balls that would otherwise head into your yard.
- jabroni_salad 4y ago
  That works great in industry, indoors. Outside you are going to have animals. I have a few acres that I mow and find new nesting animals pretty much weekly. Some of them can fly, and the migratory ones (which I do not personally know how to spot aside from a few recurring visitors) are legally protected. Sometimes they run away as they approach, sometimes they want to defend. Either way I do my best to go around and leave them their patch.
mandross 4y ago
I totally agree! Mower safety was also one of my worries once. It can be deadly even without software controlling it. A few days back in Poland a mower threw a piece of wire straight into a boy's heart (the kid is all right now). That being said, there is no way for the operator to constantly monitor what is happening around the mower. A set of sensors and software can be better at it. We can install cameras and LIDARs all around the mower and take care of safety better than humans do. There are challenges though: 1. You need to design the system in a way it fails safe. There is no explicit norm on how to do that yet. We can adapt existing norms from machine and robot safety for that purpose, but we risk missing something. 2. All the specification efforts right now are going in the direction of humans supervising software. The RC-controlled/autonomous mower is supposed to be always watched by someone. Firstly, humans suck at supervising stuff, it is a very boring and frustrating task, during which it is easy to get tired and lose focus. Secondly, people will quickly find a better task to do in the meanwhile: trim the hedge, do the detailing, do some weeding, while the mower is doing fine :) 3. Safe vision systems are expensive. It is hard to support a 360 deg. field of view for both cameras and LIDARs at a competitive cost. There is a huge temptation to reduce the BOM cost to compete with human labor. Now it is time for a shameless plug: personally, I am involved in making an autonomy kit for z-turn mowers. It is not an easy task.
- matt_attack 4y ago
  Yeah, losing focus while a robot drives is impressively easy even if you know you need to watch for safety. For some reason it's just so easy to trust them once they work even for just a little bit. It's part of the reason "level 3" robotic vehicles won't pan out. Can also confirm about the zero turn mowers. Recently switched jobs from automating cars/tanks to mowers myself. At the very minimum mowers have way tighter power and space constraints plus hydros are not easy to control well automatically.

I love the idea of this because I hate lawn maintenance. But the idea of letting software I wrote run around with whirling blades of death underneath of it would keep me from doing it.

vl 4y ago

They already make robotic lawn mowers. You can go and buy one right now.

Nextgrid 4y ago

A solution could be to just secure the area well enough to guarantee that 1) an uncontrolled machine can't escape and 2) people don't enter the area. Then you don't particularly care if the machine is "unsafe" if there's no way it can harm anyone. That's the approach most industrial robots take - instead of trying to deal with humans and not hurt them, just make sure there are no humans within reach of the robots.

stevehawk 4y ago
> 2) people don't enter the area That's harder than it should be. No amount of asking my neighbors has kept their kids out of my fenced lawn to retrieve their soccer balls. Which is actually an issue because one of my dogs is not particularly friendly. I'd gladly kick the ball back over if they just knocked on the door. We're not a neighbor-hate-neighbor situation, we actually get along quite well, but kids will be kids.
- Nextgrid 4y ago
  Asking isn't enough, but proper fences should both make it impossible to get in without significant effort and also do a good job at containing the machine should it try to run away. It's up to you to run the numbers whether the cost of securing the area is less than what you spend (in terms of time/inconvenience/etc) manually mowing the lawn.
  - scarby2 4y ago
    Coming from the UK I'm always amazed that many (suburban) Americans don't have 6-8 ft fences surrounding their back yard.
    - Nextgrid 4y ago
      On the other hand, if everyone's yard is accessible, it's unlikely that yours will be the one subject to anything bad so I guess it's a number's game.
    - stevehawk 4y ago
      most of us outside of major cities or suburbia would lose our views.
- ericd 4y ago
  An 8 foot mesh deer fence should do the trick ;-) it’ll also rebound lots of those balls that would otherwise head into your yard.
jabroni_salad 4y ago
That works great in industry, indoors. Outside you are going to have animals. I have a few acres that I mow and find new nesting animals pretty much weekly. Some of them can fly, and the migratory ones (which I do not personally know how to spot aside from a few recurring visitors) are legally protected. Sometimes they run away as they approach, sometimes they want to defend. Either way I do my best to go around and leave them their patch.

mandross 4y ago

I totally agree! Mower safety was also one of my worries once. It can be deadly even without software controlling it. A few days back in Poland a mower threw a piece of wire straight into a boy's heart (the kid is all right now). That being said, there is no way for the operator to constantly monitor what is happening around the mower. A set of sensors and software can be better at it. We can install cameras and LIDARs all around the mower and take care of safety better than humans do. There are challenges though: 1. You need to design the system in a way it fails safe. There is no explicit norm on how to do that yet. We can adapt existing norms from machine and robot safety for that purpose, but we risk missing something. 2. All the specification efforts right now are going in the direction of humans supervising software. The RC-controlled/autonomous mower is supposed to be always watched by someone. Firstly, humans suck at supervising stuff, it is a very boring and frustrating task, during which it is easy to get tired and lose focus. Secondly, people will quickly find a better task to do in the meanwhile: trim the hedge, do the detailing, do some weeding, while the mower is doing fine :) 3. Safe vision systems are expensive. It is hard to support a 360 deg. field of view for both cameras and LIDARs at a competitive cost. There is a huge temptation to reduce the BOM cost to compete with human labor. Now it is time for a shameless plug: personally, I am involved in making an autonomy kit for z-turn mowers. It is not an easy task.

matt_attack 4y ago
Yeah, losing focus while a robot drives is impressively easy even if you know you need to watch for safety. For some reason it's just so easy to trust them once they work even for just a little bit. It's part of the reason "level 3" robotic vehicles won't pan out. Can also confirm about the zero turn mowers. Recently switched jobs from automating cars/tanks to mowers myself. At the very minimum mowers have way tighter power and space constraints plus hydros are not easy to control well automatically.