Bird Bot

Once upon a time, there was a bird-owning engineer with a problem: his African Grey Parrot would, like most of their species, become very upset when away from what they considered the center of action in an area, and would start hooting and screeching for attention. His first solution was a noise-detecting squirt gun. At first this worked, and the spray of water would interrupt the screaming. Then the bird figured out how it worked and started intentionally triggering it for a quick bathtime, which parrots love. (African Greys have the intellectual capacity of a 3 year old on average, but have some skills that human children don't figure out until they're 12.)

It's not safe to let a parrot walk around unattended, one because they are immensely small and light and they will die if you step on them. (A medium sized parrot such as an African Grey weighs about one pound. The heaviest known parrot weighs 8 pounds.) Another reason is that they can chew on things on the floor, or even the floor itself, causing immense property damage. So his next idea was one that allowed the bird to travel around safely: a small motorized bird-controlled cart. The bird stands on a small perch, and pushes around a metal bar to control the cart. This way the bird could follow the humans of the house around without ever being underfoot or in the path of tempting electrical chords, rugs, or floor tiles.

The joystick component appeared to be one of the surprisingly larger engineering challenges. A parrot's beak evolved to crush nuts, is about a strong as a human with a pair of pliers, and they immensely enjoy ripping things to shreds with it. The top part can punch through wood like an awl, and the larger parrots can even destroy a steel cage. The joystick had to be designed in such a way to resist puncture, pressure, and had to endure being pulled on, all of which the bird almost assuredly tried to do.

The most challenging thing though, is that our enterprising engineer did not want to have to put this cart away every day when it's time for the bird to go to bed. The cart is designed with a computerized system that can find its way back to the charger, and slowly scoot the cart into position where it gets plugged in and charged back up for another day of bird-moving. This does so with computer-vision, which is remarkably difficult to do successfully. Also, it begins to do this the moment the parrot leaves the cart.

If I were this engineer, I would look into having this cart mass-produced.

Electric Bees

About 75% of our crops today rely on insect-based pollination.   We typically use bees for this purpose.  Bees drink nectar from the plant's flowers, getting pollen all over themselves in the process.  The pollen then fertilizes the plants, which then produce the fruit and seeds that we eat.  Everyone wins.
Except that bee colonies are in trouble.  We're not sure why, but something is killing off all the bee hives.  Some think that it is a disease not yet identified, other think that it is stray insecticidal chemicals from farms.   In any case, without bees, no fruit.
Many of these plants can be manually pollinated with a cotton swab, but that's a waste of human labor.  So instead, I think we should make robot bees.   These would fly around getting and injecting pollen, and occasionally returning to the hive for a recharge.   These robots would be immune to all diseases and chemicals, but unfortunately would not make honey.

Pistol Shrimp Bot

The pistol shrimp used to astound scientists: It did...something..., and then fish in front of it abruptly dropped dead. The fishy victim is then pulled inside the pistol shrimp's burrow and devoured. Clearly, this had to be studied.
It was discovered how this works by filming the shrimp firing in slow motion. All the shrimp does is close its claw really really fast. The rapidly increasing pressure causes a cavitation bubble with temperatures rivaling the surface of the sun. This then, surrounded by seawater, explodes slightly as everything condenses back to normal. The heat, pressure, and shock wave all injure the shrimp's prey, usually adding up to a fatality. And knowing all this gave me the idea to weaponize it.
The weaponized pistol shrimp robot would swim up to things we don't like underwater, such as the propellers of an enemy submarine. It would then bring a massive claw as close as possible as close to the propellers as possible, and then very abruptly close it. With a loud banging sound, the propellers are promptly damaged by the ensuing cavitation bubble, as cavitation is the bane of naval propellers everywhere. The submarine is now mobility-killed, and can be finished off with a cleverly placed depth charge if it continues to cause problems. Or if this is done to a surface ship, that ship isn't sailing anywhere anytime soon. If it causes more problems, it'll wind up torpedoed, or we could cavitation attack the hull to sink it.
The robot can then swim to the next target, as an enemy that has one boat probably has a lot more.
Unfortunately, this won't be too useful in modern warfare, as all our battles are mostly against asymmetric enemies who don't have navies, because they aren't nations. Most of our enemies now are insurgents, and stopping them requires a whole different type of fighting than the kind that stops nations.

Quadcopter Construction

A popular robot for fun is the Quadcopter, a robotic flying machine that has four helicopter-like rotors, and can do all sorts of aerial maneuvers by varying the speeds of its rotors. And some time ago, someone taught them to build building frameworks. Wait, what?
Discovery News reports that the University of Pennsylvania has developed Quadcopters that can manipulate plastic rods with a magnet on one end into the framework of pretty much any building. The metal end of one rod connects to the magnetic cube on the other, to form extremely solid building frames. Presumably one could finish it off with walls and floors that also attach to those magnets.
Already, people are imagining using these to throw up buildings in a hurry in places where it's impractical to take human construction workers. War zones. Mars. Antarctica. The quadcopters will cheerfully work in all of those places. And given a solar-powered charging station, they can work until they run out of parts. Admittedly, they sound like a swarm of angry bees from hell and being in the vicinity of them would be quite unpleasant, so I don't imagine them working urban construction anytime soon. (Especially because scaled up to the point where they'd make human-sized buildings, the noise would certainly rupture your eardrums.)

Brute Force Safecracking

If you wanted to get into a safe, but didn't know the combination, how would you crack it? The dumbest, but guarenteed to work, solution is to try every combination until one works. A human safecracker would get tired within a few hours of doing this, so Hack A Day reports someone automating this...with robots. The robot works faster than a human safecracker too.
The robot is a metal-and-plastic manipulator machine controlled by an embedded computer, and would fit in a backpack. If the thief is sneaky, and does this at a time when most people are asleep, and muffles the whirring noise made by the servos, he could sneak it into a bank at 2am in a backpack, muffle all noise in the area, let it grind away for 3 hours, grab the safe contents, grab the machine, and be gone by 6am. If he's stealthy enough, no one would even notice.
In some ways, I suppose this was inevitable. Cryptographic brute force is the only known way to solve NP based problems, and the only P based way to crack a safe would be to somehow figure out some pattern to the combination based on the manufacturer's serial number.

Leaf Hunting Robot

In the forest, when leaves fall to the ground, small animals tend to eat them, and if the leaf escapes that, the worms get them. Ultimately, the nutrition that went into the leaf is returned to the forest as a whole. Nature is one big cycle. We like trees too, but when those trees drop their leaves, as happens in the forest...well, the animals that would have eaten them aren't there. If nothing else, your fence is in the way. And nothing's trampling them into the ground where the worms can get them. Typically, you rake them up and throw them away, where they rot in a landfill somewhere.
I'm imagining a small robot. You active it at the beginning of fall. It wanders around your yard, looking for leaves. When it finds one, it shreds it, and buries the shreds in a convenient way. (Say, it injects it under the grass in a way that won't be readily apparent the next day.) The raking is done for you, and the worms benefit. When the worms benefit, so does the soil quality, which now has been aerated and fertilized, which the grass will appreciate. And the robot can't cost too much. You won't have to rake much. Only on the heaviest days. When the leaf fall stops, you can take your robot back into storage, or maintenance.
People's yards and spare time would both benefit.

Cockroach Hunter

Cockroaches are the blight of many an urban area. They feast on our food and garbage, and quite sturdy against many biological threats, and few things are willing to eat them. Also, they're gross and they smell bad. So, they proliferate, to our great annoyance. Usually, we try to poison or trap them, with some tricks proving more effective than others.
I'm imagining either a robot, or a specifically bred animal, that seeks out and consumes cockroaches for power. If it were a robot, it would need a bacterial digester to turn the roaches into power. (Animals have a digestive system that turns what they eat into ATP and carbohydrate chains that they can burn for energy.) If it were a robot, it would be programmed with roach-like habits, like avoiding light, and tracking pheromones. If it gets too much power, probably it can go plug itself into the wall, and save you a few cents on your electricity bill. An animal version would, if it fed well, attempt to breed, which we clearly want to encourage. Answers.com suggests that a good starting point for a professional cockroach predator would be the gecko, a small lizard with an immense hunger for insects.
I would want to make this cheap enough to drive roaches into near extinction in cities that expressed interest in this. Cockroaches would continue to survive, if they learned to avoid human settlements. (Already cockroaches know to avoid flickering lights, as this means that a human is coming with things like squishy shoes, poisons, and possibly assistant animals like cats and geckos.)

WorkCycleBot

Robots have gotten very advanced lately. Modern robots can move about, avoiding obstacles, learning where obstacles are, and can manipulate objects like a pro. This gives me an idea to improve office productivity.
A set of robots, each dedicated to one task, roams the office in a set pattern. One replaces every garbage can with a fresh one. At the end of its route, all the cans are dumped into a larger container. One passes around coffee, tea, and/or snacks. It should also take away old cups, depositing them in a washing machine at the end of the route. One delivers mail with an internal map of which worker is in which cubicle. One brings paperwork left on it to the manager's desk. And one vacuums. (That one has already been invented, and there are at least 4 brands on the market.)
With these robots, workers don't have to worry about trivial tasks, and can keep up flow for a bit longer. They will still require breaks from work occasionally, and I suspect social networking would fill the gap. If they learn any new information from that, their productivity will improve.

Pool Robots

One of the things my father loves about his house is his pool. He can go for a swim any time he feels like it, and then go right back home afterwards. No fees, no dealing with car keys, or locker rooms, just a happy dip whenever. One of the things he doesn't love about his pool is the maintenance.
Pools require a lot of maintenance. The water must be circulated, or mosquitoes will lay their eggs in it, bacteria grow and prosper, and pond scum grow, stinking up the entire yard. The water requires chemical treatment to ensure that diseased based organisms can't take root. Organic debris like leaves tend to fall in, and must be fished out. My father spends a lot of time on this. Though it will put many pool-boys out of work, I'd like to automate this work. Pools are nice, and I'd like to make them more comfortable and affordable for everyone.
Already on the market are automated machines to sweep the bottoms and sides of the pool. Experiments with automated quality control tests in factories suggest that a netting machine could work by having many cameras looking at the pool, and bringing down one of many nets on a mechanical arm to remove the offending object. A perpetually running pump circulates the water, but before it returns to the pool, it gets run through a chemical probe, which determines the numbers all pool owners need to know: pH and Chlorine levels. It would have a reserve of chemicals so that it could adjust these levels as necessary. It could also email or tweet when these chemicals are in short supply. Lastly, a trip past an ultraviolet light would kill off any bacteria or virus that survived that far. The then clean water is returned to the pool. Ideally, the pump should be underground, where the noise it makes will not be heard.
It may also be wise to have a reserve water tank. When it rains, dad's pool sometimes overflows, which annoys him. When it doesn't rain for a long time, the water level decreases. (Dad's usual solution for this is a hose.) With the reserve tank, when the water reaches a certain level, we can siphon some off into the tank, and return it later when the water level reaches a lower level. The pool would maintain a certain homeostasis. (The water tank would also need to be ultravioleted. We could probably skip on the chlorine, and mosquitoes are unlikely to lay eggs in an underground tank.)

Gutter and Sewer Bot

Civilization has a lot of tight spaces. Tight spaces which are difficult and unpleasant to clean. I'm specifically thinking of the gutter, a metal attachment to the roof that collects rain and funnels it to a specific downspout, and sewers, which takes used water away from human habitation. And by used, I mean, has waste in it. Waste that could cause disease if it hangs around too long.
Admittedly, gutters almost never have to be cleaned if you fit them with a tight grille that allows water through, but sloughs off leaves. But let's say that you don't or can't do that for some reason.
In either case, you have tight spaces, and this would best be navigated by a small, battery powered robot. With little scrubbing attachments to detach debris from the sides. It will need to be waterproof, because there's no guarantee of the area being dry, and great probability that it will be distinctly wet. The robot should scout around, loosening anything it can get its little brushes on. After a fixed but generous distance, it should return for recovery. When brought back to human hands, it deactivates, and should then be cleaned, maintained, and have its battery recharged. And you, the human owner, never left your house.
Though I understand your reservations about touching a sewer bot.

Remote Ditch Digger

A number of industries require holes in the ground. Plumbing (for water in, out), electricity, telecom, and so on. People have to come and dig. It's sweaty, sweaty work. And the pay kind of sucks.
However, telepresence gets better every day. A robotic digger sends telemetry to an office, where, on a computer, or shovel-like interface, a worker can operate the machine from air conditioned comfort. It still operates about as fast as before, but the working day is longer and more comfortable, so the low pay isn't quite so objectionable.
The machine is probably even the same shape, just with a computer and antennae instead of a human-cabin....

Autonomous Battery Exchanger

Hack a day has a fascinating system for exchanging a robot's battery, to extend its useful life. The main advantage of this is that the robot has basically no downtime, remaining on duty almost continuously. Useful, when shutting down your robot would start costing you money, as in some factories.
The mobile robot probably has a battery for a power source. Batteries are portable chemical cells that can provide energy with no cords or cables, which a robot could easily trip over. The main limitation of batteries has been that they only have so much power. Some batteries are rechargeable, and can be provided with an energy source to charge them back up again. (When provided electricity, the chemical reaction that powered them reverses, setting it back to its original state.) Non-rechargeable batteries must be thrown away when they are out of power.
This station removes the old battery with an extractor arm, and replaces it with a new one. The entire process takes 30 seconds. If the robot has a capacitor bank, it doesn't even shut down. The robot is also held in place during the process to ensure that it doesn't roll away or tip over, which would confuse its systems.
I remember a similar system described for electric cars, in which the car would drive to a station, the owner would pay with credit card, and robotic systems in the station would extract the old battery, put in a new one, and go recharge the old battery to sell it again. The batteries would, of course, have to be rechargeable and identical in shape, but car batteries already are.
So if you have a system like this with rechargeable batteries, you have the continuousness of a wall-socket, and the mobility of a battery, at the same time. A major plus for any robotics system.

Robot Leg Research

What does a three legged dog have to do with a walking robot? Plenty, says Discovery news. Both need to learn to walk all over again.
See, a dog knows how to walk with four legs, in a variety of gaits. If the dog then loses one leg (say, to an accident, injury, or whatever), the regular style of walking or running doesn't work anymore, but they can learn to limp in short order. Apparently, this encouraged a German scientist to see if this can be applied to robots, who also need to learn to walk from scratch.
Apparently, every animal on earth has an even number of legs, arranged in pairs, and roboticists have been trying to shoehorn their patterns into robots that have odd numbers of legs, and this works out rather poorly, balance wise. A pattern for an odd number of limbs would be completely different.

Mobile Animal Trap

We humans develop colonies of animals at the periphery of our civilization. Mice and Rats seek out our food stores for a free meal. Predators of mice and rats do the same for the same reason: easy food. Next thing you know, you have a very large, complex, and frankly annoying food web of animals that you constantly trip over. Also, the mice and rats have "fun" diseases that infect us. For our health, we have to nip this in the bud.
The best solution would be to quickly kill off the mice and rats, but we're squeamish. So most pest-control works by trapping. The animal is lured into one of many cages, which must be periodically checked for the presence of an animal. And when one is found, it should be removed from the cage, followed by removing it from the area. (Some animals, like skunks, are more useful elsewhere. Others, like the mice or rats, probably need to be killed off.)
I saw an episode of "Dirty Jobs," Mike Rowe's little show where he does many of the strange professions that people do for a day, and in this episode, he was a trapper, joining a professional trapper in checking all the cages. Some were empty, some had badgers, racoons, and skunks in them. The cages all had to be emptied by hand. As I was watching this, it struck me that this is fairly inefficient of the trapper's time. Therefore, it should be automated.
I have an idea of putting the cage on wheels. The cage has a small weight-trigger, and when the animal is inside it and inadvertently activates the trigger, the cage puts down the wheels and rolls itself over to a larger centralized holding area, which it would find by GPS signal. Once there, it would deposit the animal, run through a cleaning area to wash off anything the animal left behind, obtain a fresh bait, and roll over to its original location. Then the wheels would be taken back up, and the trap would be fully reset, ready for trapping a new animal.
The trapper's job would now consist of handling the animals in centralized holding, and deploying new traps. Much simpler. The decrease in workload should be offset by trapping way more areas, as the more pest control can be handled, the better.

Wallbot

Due to a recent moth infestation that we've been vigorously fighting, the walls in my house have little stains on them. This is, needless to say, annoying.
I'll probably wipe it up with cleaning sprays and towels, but I'm not the only one to have this problem. I understand that parents often have children who either inadvertently throw food at the wall, or doodle on it with crayons. So, how can I clean the walls more...automatically?
I suppose the best way I can think of is to take the "Roomba" approach and have a robot. The robot would wheel itself alone the walls, sliding up and down on metal supports, scrubbing grime from the walls, sucking the soap and absorbed dirt into a special container, and periodically leaving the walls to go dump its bin into a nearby sink.
I think people will really like this. iRobot's robotic vacuuming machine, the aformentioned Roomba, is so popular that people treat it not as an appliance, but as a pet. People will actually sweep their floors for their robot so it doesn't have to "work" as hard. Which, mind you, it is a robot. It at worst does not care, and at best could be said to actively enjoy cleaning.
You could write entire psychology books on that topic.

Togolese Recycled Robots

The Lady Ada reports that in Togo, a young man is building robots....out of old TVs.

Why? For inspiration, he says. He's demonstrating that this kind of thing is indeed possible, contrary to the expectations of the local Africans who feel that building robots is not possible in their country, which lacks the financial resources of countries where robots are more common.
Why TV's? The video doesn't say, but I'm going to guess that A) It's a complicated device with many of the necessary parts, B) When they stop working, they're thrown away in favor of buying a newer, more capable model over being repaired, C) The parts that break aren't the ones that interest him, and D) They are immensely popular, so broken TVs can be found in immense quantities.
Nice going Sam. I'd totally buy one.

Farming Afar via Game

In both the US and China, farming simulators are increasingly popular. Possibly in other countries too, I just haven't heard news reports about it. Such games involve pretending to farm by digging plots in a big grid, click for seed, click for water, click for harvest, and so on. It gives me an idea.
The games are popular because of their abstraction. What if there was a robot, able to do those various mundane tasks, with a supply of seeds and the ability to gather and distribute water? It would scan its environment, and report it to a game interface. People playing the game would indicate to "dig" various plots, water them, and which seeds to plant. The robot would receive instructions from the game, dig, water, and plant at the coordinates.
Each day it would report back to the game about the conditions. When the seeds sprouted, ground moisture and chemistry (reported as percentages of USDA recommended, "Nitrogen is at 53%, add more fertilizer!"), and the like. And players could resolve problems with a click.
At some point, the robot would report the plant ready for harvest. The player clicks. The robot slices off parts of the plant of commercial interest (like fruits, corn cobs, lettuce leaves), or if inapplicable, pulls the entire plant (carrots would be yanked, not cut), and puts it in a bin. The remainder of the plant is plowed under for fertilizer. Post-harvest, the robot reports the area as fallow, ready for another planting.
It'd be slow, compared to online farm games. There, crops are ready in as little as 2 hours, with the slowest crops taking maybe 7 days. Real crops would take at least 40 days, which may leave players feeling like they're lacking in accomplishment.
But unlike farm games, one would receive more than virtual money for this. The harvest bins would have a sale-able product, the profits from which could be split between the robot owner, the game producer, and the player. Playing the game could net you real actual money, possibly.
....nah, this is too insane.

Nairobi Robot Fair

Did you know that Nairobi, Kenya, has an annual, university sponsors, robot building competition? Neither did I.
The article doesn't explain what these robots do, but they seem interesting and useful. Most of them seem to be for the purpose of conveying small items from one place to another.

Litter Collection Bot

The city in which I live is clogged with litter. As far as I can tell, some moron gets drunk and decides that throwing crap out his car window is disposing it. Also, I occasionally find tires where somebody has decided that since disposal costs are so high, that the best thing to do with their worn out tires is to toss them where no one will notice. Someone does eventually notice, but by then the perpetrator is long gone.
Mobile robots exist that can find their way around independantly. Let's build some to roam the city, picking up all unattended objects, and throwing them into a built-in container. After a certain run, it should return home to empty the bin, and charge itself up.
Objects should be classified as either garbage or lost objects. Lost objects should have their owner determined, and be returned to their owner. Garbage should be further sorted as recyclable, burnable, or neither. Objects that are neither should be taken to landfill.
The robot should be strong enough to lift a tire, but not strong enough to lift a car. (Taking parked cars is asking for trouble.) The robot must be waterproof in case of puddles or ditches, able to move across grass, sand, and slopes up to 45 degrees, and able to rescue itself from falling into a ditch. If not rescue itself, then somehow call for human rescue. (Yes, I can imagine the robot using a cell-phone-like system to tell me in a synthesized voice that it has fallen, again, into a ditch, at such and such location, and would I please come rescue it. I can further imagine that this would be the twelth time this happened that day.)
Lastly, the robot must not wander into traffic. I doubt it would survive impact with a moving car. It must have some means of telling apart "road" from "sidewalk," and preferring "sidewalk." Bonus points if it can successfully taught to cross the street.