Slashdot Mirror
How Could Swarms of Robots Help Humanity?
ceview writes "Researchers at Sheffield Centre for Robotics have demonstrated that a swarm of 40 robots can carry out simple fetching and carrying tasks. This is done by grouping around an object and working together to push it across a surface. They believe that this could provide opportunities for us mere humans to harness such power to do all sorts of things like safety — what like catching falling workers perhaps? Youtube action here."
My father had a ranch. We used to hire 50 to 60 skinjobs to fetch tomatoes. You know it takes two people to fetch the same tomatoes now.
When the foot seeks the place of the head, the line is crossed. Know your place. Keep your place. Be a shoe.
The next time we need to hunt down someone in a cave complex we can use a drone swarm to autonomously explore all the holes. We'll only need the giant thermobaric bombs once the right caves are identified.
Yay drone swarms.
Maw! Fire up the karma burner!
>> humans to harness such power to do all sorts of things like safety — what like catching falling workers perhaps
Doubt it. However, I could see a bunch of these being loosed by a in a war zone: individually find the big metal ship, clump together beneath the waterline, go boom. Or, to clear the way for a raid, something like for the next 6 hours, find all the moving human-like shapes, get close, go boom (or clump at doors, go boom).
If a robot could be designed that could distinguish weed from crop, it would eliminate the need for most herbicides and herbicide resistant GMO seeds
I think most people here are missing the point. I doubt the aim of his research is to develop things that are immediately useful. It is more about understanding the behaviour of complex multi agent systems. A lot of systems around us follow this same model. For example, the economy is just the sum of the actions of many "simple" agents. This research aims to look into exactly that. Really it is amazing that few simple agents without any higher form of control can accomplish anything when their behaviour is based on maybe 2-3 rules.
The goals of this research are very simple and very controlled. Take for example the pushing of the box. It was one box in a pristine room with one target. The pushing algorithm is extremely simple. Move randomly until you can not see red and then move toward the last location that you saw red.
What happens if there is a small obstacle on the floor and the object can not be pushed straight toward the floor?
What happens if the targets not visible from the object?
What happens if there are several possible targets and several possible objects to push?
How do the robots know when to stop pushing?
Sure simple robots with simple programming given simple tasks in a simple environment will work. The real world is rarely that simple. As one throws more real world parameters into the situation the complexity rises exponentially.
They believe that this could provide opportunities for us mere humans to harness such power to do all sorts of things like safety — what like catching falling workers perhaps?
Is that worker in an uncontrolled fall or are they jumping? When you add decision making into a process it becomes much more complex than pushing or sorting.
Pick up trash.
Mow lawns.
Sort trash to determine recyclables.
Fix potholes.
Paint over graffiti.
Repair underground infrastructure - gas and water pipes.
Detect leaks in underground infrastructure.
Remediate contaminated soil.
So many options.
- Zav - Imagine a Beowulf cluster of insensitive clods...