Where are all the robots?  In my early career I had expected that behavior-based programming, the idea that made Roomba possible, would quickly lead to an avalanche of other robot applications.  That hasn’t happened.  And even in the case of Roomba, 16 years passed between the publication of the new paradigm and Roomba’s launch in 2002.  I chalk up that delay to a failure of imagination.  These days the number of robot applications earning their spot in the marketplace is increasing but, to me as least, it still seems distressingly slow.  Below is an incomplete list of new robot tasks that I’d like to see.   Some may be years away, but—given sufficient imagination—I think most are within reach now.  They await, like Roomba did, the right spark of inspiration.

1.   Mow My Lawn

I have a horrible lawn.  It contains exposed tree roots, a steep slope, ruts, bare spots, unfenced flower beds, and boulders of all sizes left as parting gifts by an ice age glacier.  I’ve been mowing that grass myself for over 30 years and each time I do I think again about having a robot take over for me.  Sadly, I haven’t seen a commercially available robot mower that appears equal to the task.

When designing a robot lawn mower, one naturally first imagines a lawn worthy of Better Homes & Gardens.  A lawn that’s lush and flat; one that possesses none of the edge cases that make robot navigation tough.  That sort of model lawn would make my grounds green with envy—well a sort of splotched, mottled green.

It may be the edge cases—the many hard-to-recognize, difficult-to-deal-with hazards that less-than-ideal lawns like mine exhibit—that are delaying the rapid adoption of robot mowers in the US.  

There’s another irksome issue with most existing robot mower.  Typically there’s a gap between the mowing blade and the edge of the robot.  The gap necessarily leaves a strip of uncut grass around obstacles like fences and the trunks of trees.  Another source of uncut grass is the safety margin the robot must leave to avoid either driving onto flower beds or dropping a wheel into the sharply cut step edge that often outlines flower beds on well-maintained lawns.  Regardless of the cause, eventually the owner must manually trim that unkempt strip with a weed whacker or other implement.   

Roomba had a similar problem; the main brush could not extend all the way to the edge of the robot.  We solved the dirty margin problem by adding a side brush that swept otherwise unreachable dirt into the path of the main brush.  But an analogous solution doesn’t seem possible for a robot mower.

I’m sure that existing robot lawn mowers work just fine on many lawns.  But I’m still waiting for a mower that can tame a lawn as wild as mine.

2.   Rake My Leaves

An annual battle with leaves commenced years ago when I moved to suburbia.  My rustic yard was home to an abundance of trees and each fall they rained down their colorful foliage.  Over time I tried different techniques to corral the discarded vegetation and I measured the efficiency of each new attempt.  Nothing stood out, every method I tested took around 24 total hours of work.  As I toiled, I thought about how a robot might do the task or at least help me with it.  Sadly, I came up with nothing that checked all the relevant boxes.

Collecting (not necessarily raking) leaves is a more difficult task for a robot than mowing.  The technical challenges are numerous.  Does the robot need to identify where the leaves are, or can it just move about while applying the collection mechanism to the ground?  How does the robot collect the leaves without damaging the lawn or flowers and other desired plants?  Presumably the robot delivers leaves to some designated point, but as the pile grows the point must change—how should that work?  Finally, how do you achieve all that technical sophistication at a price attractive to customers when they are likely to use the product only one or two weeks out of the year?

3.   Snow Blow My Driveway

A number of companies either have tried or are trying to build robotic snow blowers.  Living in New England I’ve often been the recipient of requests for such a device.  Several features of the task make it difficult for a robot.

First, even in my part of the country, a snow clearing robot will spend the great majority of its days sitting idle in the garage.  Economics are challenging for expensive products that are used rarely and unpredictably.  It’s worse for a robot snowblower than a manual one because the robot is likely to cost more.  

How does the robot know the boundaries of its operational area and where is it OK to place the snow?  It must know both things with great reliability and it must be easy for the user to convey this information to the robot.  

Should the robot wait until the storm has ended before starting to clear the snow or should it clear continuously?  In the former case, when many inches of snow may blanket the ground, the robot must be large, powerful, and fairly expensive.  This raises the barrier to entry for new customers.  In the latter case the robot can be smaller and less expensive making it more attractive to customers.  But if it isn’t deployed soon enough after the storm begins, the snow may become so deep that the robot can’t be used at all.

The problems don’t appear insurmountable.  But I know of no company that has truly  threaded this particular needle.

4.   Paint My House

For years robots have been painting cars and other mass-produced items.   It’s fun to imagine using robots to accomplish the same feat for houses.   Robot painting might enable any house to be painted as elaborately as desired.  Certainly, I’d have my house repainted in a timelier fashion were the process less costly and less disruptive.

In the factory, robot painting often involves a conveyor that carries the target item past one or more manipulator-type robots, enabling the robots to reach all parts of the item.  Simply scaling such equipment up to the size of house seems problematic.  Also, part of the appeal is that houses have trim and other features that should be painted in different colors.  In factories, generally only one color is sprayed on robotically.

It isn’t clear how the painting process should be reimagined to make it both possible and economically advantageous to have robots do the work.  But I’m hoping for a solution—my house could use another coat of paint.

5.   Clean My Bathroom

After Roomba became successful, customers would often approach iRobot with requests for other robots.  A frequent one went like this:  “You built a robot that cleans my floors, so how about one that cleans my toilet or bathtub?”

Of course, the engineers all knew that the technology that enabled floor cleaning was thoroughly unrelated to the technology needed for other sorts of cleaning tasks.   Roomba would simply fail if asked to clean a surface that wasn’t mostly flat and horizontal.  And there were no simple tricks that would let us easily expand its scope.

So with knowing smiles we deflected such requests.  But it made me feel uncomfortable to inform customers that they were asking for the moon.  Taking over disagreeable chores like cleaning the bathroom is exactly the sort of thing robots should be doing.  It was our limited imagination that was the problem, not the customer’s need.

I’ve seen attempts to build automatic bathroom cleaners.  SC Johnson makes a simple automatic sprayer that aims to clean the shower—relying mostly on chemistry, the company’s specialty.  But the scope is limited.  Somatic, https://getsomatic.com, promises a robot that cleans the entire bathroom, but it’s designed for commercial cleaning and, even if it works, it wouldn’t be appropriate for my bathroom.  

Neither approach quite hits the mark for me.  Some creative rethinking of the task would seem to be in order.  I have an unproven idea that I’d like to develop further, but I need to go clean my bathroom.   

6.   Keep Track of My Stuff

I’m a pack rat.  I have trouble discarding things, always imagining that someday they will be useful.  Occasionally, I’m actually right.  But by then the perfect item that has waited so long for its moment in  the spotlight is unlocatable.  Lost somewhere amid the general clutter of my office or maybe it’s in the basement or possibly the attic.

I’ve often imagined that something I call the “Personal Warehouse Robot,” PWR, could solve my problem. (Technology is so much easier than self-discipline!)   My PWR would enable me to have an uncluttered and stylish home, tastefully populated with objets d’art.  Yet, every quaint item that I’d ever hidden away would be instantly accessible—the PWR would unfailingly retrieve it from its hidden storage area.  The robot might be similar to the Kiva System’s robot (now Amazon Robotics) that brings items to an order picker.

How to implement such a device in any practical way has so far escaped me.  But if someone wants to build one at an affordable price, I’ll buy it!

7.   Banish Kudzu

One of the most useful things robots might do for the environment is to control invasive species.  Problems occur when a robust species is introduced into an area where it has no natural enemies.  There it can overwhelm native species and generally wreak havoc.  So, if a problem species has no natural enemies, why not build one?

It’s not hard to imagine swarms of robots prowling the woods or the sea, ever vigilant for a plant or animal that doesn’t belong.  At least one or two attempts to build robots that can hunt Lionfish, an invasive species in the Atlantic, have begun.  Much more seems possible.  

The economics of the enterprise is another matter.  A sufficiently inexpensive underwater robot designed to harvest Lionfish might be able to pay for itself by delivering its catch to a market.  But robots intended to control other invaders might have to be financed through taxes or in other ways.  Surprisingly, there is a market for Kudzu, but at least in the US, it’s quite small. 

8.   Find Lost Hikers

It’s always tragic when someone goes missing in the woods, the mountains, or at sea.  Typically, aircraft with eagle-eyed observers are sent out to scan vast areas for any sign of the missing.  But having only a few searchers who must fly low enough to spot say, a person’s head bobbing in the water, means that the search is likely to be slow and difficult.

How much more efficient might it be to send out dozens or hundreds of low-flying robotic aircraft that could search at a much lower altitude?  Current systems are optimized for people.  This may mean large aircraft supplemented with sensitive sensor systems that can operate at long range.  But a reimagined system, designed for low altitude drones might offer better performance.  Maybe, rather than being designed to remain aloft for many hours, when their energy is exhausted the drones simply land on and float in the sea, waiting to be retrieved and refueled later.

9.   Pick Up Trash

There is never too little litter.  Always parks, and beaches, and highways seem to have too much.  Having people pick it up is an expensive chore.  But the task seems ideally suited for a robot.  The robot could live in the park or along the side of the road and could be powered by the sun.  It would spend its days alternately charging and patrolling for litter.  Natural beauty restored!

Several challenges stand in the way.  Can the robot reliably identify what is litter and what is not?  AI has achieved amazing results in the field of image classification but the variety of debris and of things that might look like debris to a robot is also amazing.  Once identified, how does the robot grab the litter?  Debris comes in a great variety of sizes and compositions, so will a simple two-fingered gripper suffice?  What sort of arm is needed?  To minimize cost the arm should have as few degrees of freedom as possible.  But it must also be versatile and rugged.  How does the robot release the debris?  Sticky debris may not drop off just because the gripper opens.  How is the debris stored on the robot and where and how is it offloaded?  

But the toughest challenge of all is likely to be creating a solution that’s truly less expensive than hiring someone to do the work. 

10. Deal with Laundry

Generally, I carelessly throw my clothes on the floor at bedtime.  In the morning I take clean clothes out of the dresser or closet.  In my ideal world, that would be the extent of my interaction with the laundry—everything else would happen invisibly. 

Some proposals for washing, folding, and storing clothes envision a humanoid robot doing all these steps almost exactly the way a human would do them.  But that solution holds no appeal for me—I don’t want the equivalent of a houseguest who never leaves taking up space and trundling about my home.  

Robotic apparel management, RAM, might be done in a completely different way.  The tricky issue is always: What part of the task do you hold constant and what part do you allow to vary?  When designing robots, I like the freedom of being able to change pretty much everything.  For example, does there have to be a large washer and dryer in the basement that cleans and dries bulk loads of clothes?  Maybe there’s a different, equally effective way to clean them one at a time if a robot is involved.  Do the clothes have to be picked up from the floor?  I’m willing to dump my duds in a bin if it would make the rest of the task practical.  Do garments have to be folded and stacked in a drawer?  Maybe the laundry system could be integrated with the Personal Warehouse Robot.  The dresser and armoire could then disappear from the bedroom to be replaced by a small cabinet that holds only my ensemble for that day.  Either I’d choose it, or even better for me, the powering AI would select the perfect combo.  The part of the task that is constant is that clean clothes are always available.  Everything else can be changed to make RAM economically feasible.  

* * *

The ideas sketched above are neither all mine nor all new.  Many of these notions have been around for years.  They share that trait with Roomba.  Autonomous floor cleaners had been envisioned for a century before Roomba made its debut.  The challenge is always to build the thing we want using only the tools and materials we have at the time.  A liberal application of ingenuity and innovation to one of the problems above may make it possible today rather than waiting until further research and development make it easier tomorrow.