> A robot needs more material than just recycled plastic (glass/metals? power source? processing? fasteners/glue/resin to stick it all together? Plastic-collection device?).
The bulk of the robots will be very simple bio-mimetic structures made out of bubbles of plastic attached to each other with glue (in situ production of a suitable marine glue from e.g. algae is one very big open question for the project. Although I can make do with filaments and knots.) Actuators will be mostly simple pneumatic/hydraulic systems. Peristalsis. Power will mostly be passive scavenged from the environment. MSO is an exothermic reaction. Think of giant artificial hydras lining a vast floating spiral "digestive system" that concentrates plastic at the processing center at the, er, center, while rejecting fish and other biota.
> Undoubtedly, there will be an addition of material to the ocean system.
Yeah, the limiting factor will likely be small robust electronics packages. Still, not every drone has to have the full complement of sensors and processors. It's not impossible that a simple "clockwork" brain will suffice for the bulk of the drones.
> Will the all-in cost of adding the additional material offset the all-in cost of the amount of plastic reduction?
I figure so, because I'll be only adding at much matter as needed to make the trash "smart". And the ratio of stuff removed to stuff added should be 10,000-to-1 or better. One ATMega microcontroller can provide brains for several tons of plastic.
> What damage can these devices do when they malfunction? What environmental harm can happen when they break?
Very good questions. So far I'm designing them to be no more of a problem than the trash itself. I expect to be able to have a boat pass through and neither it nor the swarm should even notice each other, FWIW. Also, the robots themselves will be recycled once they have passed the inflection point.
The main issue was how to maintain control, and I think I've got that figured out. I had to design a system that lets me specify my hardware and software from the gates up to be provably correct. Now that I have that the actual OS and robot guidance system is pretty straightforward. It helps that things like ML have come a long way since I started. A lot of problems I anticipated have already been solved in the the meantime.
> Is this even a good approach? What part of the problem does this tackle?
When I first heard about the Great Pacific Garbage Patch I was horrified, but then I looked at it from the POV of Permaculture, wherein you see this sort of thing as a resource, rather than a problem. As a resource, the plastic waste represents a huge trove of carbon atoms, they are just in the wrong place. (They also aren't bio-compatible, but that's a whole 'nother story.) Using TPD or MSO the trash becomes a resource.
> It seems like the main problem is "How do we get the most plastic out of the ocean possible". The "Many Robots" idea seems to identify the least difficult of the problems, which is "What do we do with the material once it is collected".
The swarm is designed for the collection phase.
The recycling into more robots is just to let a small initial system scale to handle a global problem in a reasonable amount of time.
> Indeed, I suspect 1,000 super durable metal robots made on land that have sufficient technology would be superior to 10,000 'recycled' bots. Indeed, 1 gargantuan vehicle cruising the ocean and pumping out giant plastic floating cubes would be even more efficient.
Efficiency is less important that it might seem, because there are plentiful ambient energy differentials on the ocean. The issue is range and detail: you have to reach most of the oceans down to the molecular level to really clean this mess up. I'm not good at enlisting the aid of others so in order to tackle this problem I had to design a system that scales with relatively little human inputs. I can't build a "gargantuan vehicle" directly but I can build a small system that can replicate itself (with a little help) and scale that up. Eventually I will have gargantuan vehicles.
> Then again, if you've made a robot that can collect micro plastics from the ocean and reproduce itself well enough such that its prodigy can do the same, then you've already won.
The bulk of the robots will be very simple bio-mimetic structures made out of bubbles of plastic attached to each other with glue (in situ production of a suitable marine glue from e.g. algae is one very big open question for the project. Although I can make do with filaments and knots.) Actuators will be mostly simple pneumatic/hydraulic systems. Peristalsis. Power will mostly be passive scavenged from the environment. MSO is an exothermic reaction. Think of giant artificial hydras lining a vast floating spiral "digestive system" that concentrates plastic at the processing center at the, er, center, while rejecting fish and other biota.
> Undoubtedly, there will be an addition of material to the ocean system.
Yeah, the limiting factor will likely be small robust electronics packages. Still, not every drone has to have the full complement of sensors and processors. It's not impossible that a simple "clockwork" brain will suffice for the bulk of the drones.
> Will the all-in cost of adding the additional material offset the all-in cost of the amount of plastic reduction?
I figure so, because I'll be only adding at much matter as needed to make the trash "smart". And the ratio of stuff removed to stuff added should be 10,000-to-1 or better. One ATMega microcontroller can provide brains for several tons of plastic.
> What damage can these devices do when they malfunction? What environmental harm can happen when they break?
Very good questions. So far I'm designing them to be no more of a problem than the trash itself. I expect to be able to have a boat pass through and neither it nor the swarm should even notice each other, FWIW. Also, the robots themselves will be recycled once they have passed the inflection point.
The main issue was how to maintain control, and I think I've got that figured out. I had to design a system that lets me specify my hardware and software from the gates up to be provably correct. Now that I have that the actual OS and robot guidance system is pretty straightforward. It helps that things like ML have come a long way since I started. A lot of problems I anticipated have already been solved in the the meantime.
> Is this even a good approach? What part of the problem does this tackle?
When I first heard about the Great Pacific Garbage Patch I was horrified, but then I looked at it from the POV of Permaculture, wherein you see this sort of thing as a resource, rather than a problem. As a resource, the plastic waste represents a huge trove of carbon atoms, they are just in the wrong place. (They also aren't bio-compatible, but that's a whole 'nother story.) Using TPD or MSO the trash becomes a resource.
> It seems like the main problem is "How do we get the most plastic out of the ocean possible". The "Many Robots" idea seems to identify the least difficult of the problems, which is "What do we do with the material once it is collected".
The swarm is designed for the collection phase.
The recycling into more robots is just to let a small initial system scale to handle a global problem in a reasonable amount of time.
> Indeed, I suspect 1,000 super durable metal robots made on land that have sufficient technology would be superior to 10,000 'recycled' bots. Indeed, 1 gargantuan vehicle cruising the ocean and pumping out giant plastic floating cubes would be even more efficient.
Efficiency is less important that it might seem, because there are plentiful ambient energy differentials on the ocean. The issue is range and detail: you have to reach most of the oceans down to the molecular level to really clean this mess up. I'm not good at enlisting the aid of others so in order to tackle this problem I had to design a system that scales with relatively little human inputs. I can't build a "gargantuan vehicle" directly but I can build a small system that can replicate itself (with a little help) and scale that up. Eventually I will have gargantuan vehicles.
> Then again, if you've made a robot that can collect micro plastics from the ocean and reproduce itself well enough such that its prodigy can do the same, then you've already won.
Riiiiiiight? ;-)