Baxter and US Manufacturing

18 posts

Flag Post

I realise I still have one open robotics thread, which I will get back to, as soon as I can satisfactorily answer some of the questions posed by other posters.

However, I encountered something in the news over the last few days which looked interesting. Then I did some digging into the subject, and was left with more questions than answers.

There is a US company called Rethink Robotics, based in Boston. They are trying to bring manufacturing back to the US. Their way of doing this, is with a robotic worker called Baxter, who is cheap, can be shown how to do a task visually, and will learn on the job how to do it best. Baxter is human-safe, and can work in a mixed environment with humans, as opposed to working inside a safety cage.

Baxter uses facial expressions to communicate with co-workers, and responds to gestures in being shown how to do a task.

Now, the beauty of Baxter is that it can work day in, day out without stopping, and does not draw a wage. It is cheaper to buy a baxter than to employ a human for the same position, and support is part of the contract. If used en-masse, it will definitely bring manufacturing back to the US, as these robots can undercut even most third world countries as far as wages are concerned, and bring prices down to rock-bottom low.

However, this means our manufacturing process will be largely automated, and if successful, even as the number of manufacturing industries in the US increases, the number of jobs in manufacturing outside of management, is likely to go down, not up.

Is such a trend a good thing for the US? Is it better to attract all the industries we can, even if more jobs will not be created, or should maximising jobs not industry be our primary focus as a nation?

 
Flag Post

It’s definitely an interesting idea for the future. At this point though, I doubt it would be cost-effective over standard robotics used for manufacturing or have the same capabilities as a specially designed/programmed robot. However, without studying the numbers or seeing Baxter in action, I can’t say for certain.

As far as automation – manufacturing still gets more automated by the day, although normally by standard robots. (ABB, FANUC, etc.) People forget though that for every job eaten by robotics, someone designed the robot, programmed the robot, maintained the robot, etc. So personally, I don’t see ‘humanized’ robots as much of a catalyst for anything, as I believe the ‘human’ aspect is more a gimmick than anything, as of now at least.

 
Flag Post

I think that once robots become advanced enough to begin increasing unemployment the only thing we can do is increasing taxes world wide and taxing the factory owners to pay money to the now new unemployed.

 
Flag Post

Once the whole industrial sector becomes automated we can finally do away with money and handle stuff the way they do it in Star Trek. *dreams away*

 
Flag Post

Well, from what I know, robots don’t respond well to unforeseen circumstances, and can’t optimize processes on the go like people do. And automation tends to slow down production, even if it makes it cheaper. Which means people are still very important in production and that won’t change for a while.
As for the industry vs. jobs question; any automated system needs LOTS of maintenance. By people. And those are the jobs for the newly unemployed. It is simply a matter of foresight: if you see that the industry is increasingly using robots, you plan training and education accordingly. Factories don’t get automated overnight.

 
Flag Post
Originally posted by Suicidal_waffles:

Well, from what I know, robots don’t respond well to unforeseen circumstances, and can’t optimize processes on the go like people do.

Which isn’t really relevant, since industrial robots are made for one purpose and nothing else and the production gets adjusted to that for example by introducing different production stages: welding, painting etc.
They do not need to respond to unforeseen circumstances.

And automation tends to slow down production, even if it makes it cheaper.

Find me a reliable source on that, written by someone who knows that the industrialisation has happened and I’ll make you a sandwich and send it to you via air mail.

 
Flag Post

Unless I’ve missed something here, I’m lining up with BobTheCoolGuy on this one. In what way is Baxter anything more than a slightly more sophisticated version of what we already have? And with a payload of 5lb per arm, this machine is quite limited in its applications.

We’ve had automation since the Jacquard loom, but it did nothing to impede the industrial boom of the nineteenth century. However goods are manufactured, they bring trade, and that is always good for the economy. Somebody may establish a factory where the shop floor is entirely automated, but it still needs management staff. They in turn require a whole raft of support industries – office equipment, communications, maintenance staff, dinner ladies, cleaners etc. The factory still needs packers, warehousemen, delivery drivers and so on. The finished goods might be exported, bringing in money from around the globe. What’s not to like about that?

In the long term, trying to maximise jobs at any cost doesn’t work. We tried that decades ago, and ended up paying a workforce to produce goods that nobody wanted to buy, notably in coal mining, steel, the motor industry and ship building. We just couldn’t compete on price with developing nations. Some of these industries were propped up by the government for years, and in the end the whole house of cards came tumbling down, causing widespread deprivation and social unrest. With the benefit of hindsight, it would have been far better to let them to run down gradually, allowing new and more modern enterprises to take up the slack in the workforce. So I would say bring it on. A lot of workers may have to learn new skills, but robots are going to create new jobs somewhere down the line.

 
Flag Post
Originally posted by EPR89:
Originally posted by Suicidal_waffles:

Well, from what I know, robots don’t respond well to unforeseen circumstances, and can’t optimize processes on the go like people do.

Which isn’t really relevant, since industrial robots are made for one purpose and nothing else and the production gets adjusted to that for example by introducing different production stages: welding, painting etc.
They do not need to respond to unforeseen circumstances.

SOMEONE has to respond to an unforeseen circumstance, which is why you can’t replace all your workers with Baxters. In a paint manufacturing facility, a human would notice if suddenly the paint was the wrong color. A robot would not unless you managed to specifically program that – but at that point it’s not really “unforeseen”. Maybe the machinery is making some really strange noises. Again, the robot won’t even notice unless it was programmed to – and it seems to me that “weird noises” would be something difficult to program for. Maybe the fire alarm just went off. Or an earthquake warning. Not that humans will always do the right thing in a panic situation, but the robots almost certainly won’t unless you programmed them in advance. The robot is almost certainly not going to shout a warning or push you out of the way if something big and heavy is about to land on you.

I remember once I was working in a cheese making facility for a summer job (Hey, I’m from Wisconsin) and a worker found a razor blade in one of the cheese blocks. Now, nobody is trained on what to do in a situation like that – but the worker knew enough to stop the line so the manager could figure out what went wrong. A robot would only stop the line if you programmed him to do so. If you were lucky, it would see something was wrong, mark that particular package as “bad”, and move on. If you were unlucky, it wouldn’t see the significance of something so thin and let it pass.

 
Flag Post

I recently saw a system that might help out with the whole “unexpected situation” problem. Basically it was a system that detected anything out of the ordinary and tries to match it with a category then it reports the problem to a supervisor. As most factories only run into such problems a few times a day one worker can keep an eye on several factories. So instead of several thousands of jobs you only have a handful.

 
Flag Post
Originally posted by BobTheCoolGuy:

It’s definitely an interesting idea for the future. At this point though, I doubt it would be cost-effective over standard robotics used for manufacturing or have the same capabilities as a specially designed/programmed robot. However, without studying the numbers or seeing Baxter in action, I can’t say for certain.

This one in particular, is designed to be programmed by gesture. It mimics your movements, and then uses its own AI to vary those gestures as it needs to, to perform it’s alloted tasks. There is also an API if you wish to add programming directly.

People forget though that for every job eaten by robotics, someone designed the robot, programmed the robot, maintained the robot, etc.

A possible concern is, less jobs are created than the robot replaces. If you buy fifty of these for example, you aren’t going to need fifty people to maintain them, when you can cover them with five. Assembling them is also likely done by robot. So there is a net loss of jobs there.


Originally posted by thijser:

I think that once robots become advanced enough to begin increasing unemployment the only thing we can do is increasing taxes world wide and taxing the factory owners to pay money to the now new unemployed.

Well there are many possible solutions, but all require something of a paradigm shift from the way we do things now. With quite a conservative country, that may cause quite substantial problems.


Originally posted by EPR89:

Once the whole industrial sector becomes automated we can finally do away with money and handle stuff the way they do it in Star Trek. *dreams away*

That’s a long way away however. We’re not ready to do away with money as things stand. The terrain on the route towards that destination is also…somewhat inhospitable.


Originally posted by Suicidal_waffles:

Well, from what I know, robots don’t respond well to unforeseen circumstances, and can’t optimize processes on the go like people do.

Depends whether the controlling AI is strong or weak. This seems to be a weak AI, with elements of a strong thrown in. I would have to see how it handles conflicting orders or parts that are different from what it expects. I suspect Baxter will respond with confusion, and call for assistance.

And automation tends to slow down production, even if it makes it cheaper.

Part of the value of humanoid, multi-purpose robots is they don’t slow down production. In fact by allowing the robot to work right next to humans, you shouldn’t see any decrease in speed at all.

As for the industry vs. jobs question; any automated system needs LOTS of maintenance. By people. And those are the jobs for the newly unemployed. It is simply a matter of foresight: if you see that the industry is increasingly using robots, you plan training and education accordingly. Factories don’t get automated overnight.

Not as many jobs as you might think. But yes, retraining is mandatory. However, most aren’t going to look ahead, andare not going to be capable of such training in the first place. What do you do with these large numbers of increasingly unemployed, unskilled workers? Is that not a recipe for social unrest?


Originally posted by beauval:

Unless I’ve missed something here, I’m lining up with BobTheCoolGuy on this one. In what way is Baxter anything more than a slightly more sophisticated version of what we already have?

The main differences are a self-learning loop, where Baxter gains more experience as it performs a task more often, and the ability for it to work safely in a mixed environment without any need for a safety cage to protectthe human workers. You can reach right over it, and still be perfectly safe.

And with a payload of 5lb per arm, this machine is quite limited in its applications.

True, but that’s a this-generation issue.

But it still needs management staff. They in turn require a whole raft of support industries – office equipment, communications, maintenance staff, dinner ladies, cleaners etc. The factory still needs packers, warehousemen, delivery drivers and so on. The finished goods might be exported, bringing in money from around the globe. What’s not to like about that?

Well, warehouse cobots have been around for a while, and are quite commonly used now, so warehouse staff can be safely halved. Robotic cleaners are already available commercially, and bring the same benefits as robotic line workers – no wages, utter dedication to the job etcetera.

Delivery drivers will likely start being automated within the decade. We already have some self-driving cars on US roads. Maintenance staff will still be human, as will the facility nurse, and the upper management. But as the total staff count decreases, so does the need for cafeteria staff, and things like HR and payroll start to downsize.

It’ll be great for the economy, but long-term, blue collar jobs will increasingly disappear. It is whether or not this is going to be a problem for our society.

In the long term, trying to maximise jobs at any cost doesn’t work. We tried that decades ago, and ended up paying a workforce to produce goods that nobody wanted to buy, notably in coal mining, steel, the motor industry and ship building. We just couldn’t compete on price with developing nations. Some of these industries were propped up by the government for years, and in the end the whole house of cards came tumbling down, causing widespread deprivation and social unrest. With the benefit of hindsight, it would have been far better to let them to run down gradually, allowing new and more modern enterprises to take up the slack in the workforce. So I would say bring it on. A lot of workers may have to learn new skills, but robots are going to create new jobs somewhere down the line.

Mmm, so would government re-education programs perhaps be the best way forwards in dealing with such increasing robotisation? Find jobs for these people in other sectors,such as retail where human-human interaction is still absolutely vital, or clerical work which is difficult to automate?


Originally posted by Ceasar:

SOMEONE has to respond to an unforeseen circumstance, which is why you can’t replace all your workers with Baxters. In a paint manufacturing facility, a human would notice if suddenly the paint was the wrong color.

A machine vision system should pick up on that as well.

Maybe the machinery is making some really strange noises. Again, the robot won’t even notice unless it was programmed to – and it seems to me that “weird noises” would be something difficult to program for.
Maintenance staff will be needed, yes.
Maybe the fire alarm just went off. Or an earthquake warning.

A fixed robotic system won’t need to evacuate the building. It can stay in situ. Insurance covers the cost of replacing hardware, and the mind of the bot is backed up. Just plug the data into the new ones.

The robot is almost certainly not going to shout a warning or push you out of the way if something big and heavy is about to land on you.

True, but in a heavily automated facility, wouldn’t that be an argument for more automation, not less? Health and safety concerns dictating that less humans should be in the area.

I remember once I was working in a cheese making facility for a summer job (Hey, I’m from Wisconsin) and a worker found a razor blade in one of the cheese blocks. Now, nobody is trained on what to do in a situation like that – but the worker knew enough to stop the line so the manager could figure out what went wrong. A robot would only stop the line if you programmed him to do so. If you were lucky, it would see something was wrong, mark that particular package as “bad”, and move on.

From what I understand of Baxter, that is what would happen in this situation. It would be confused by the item being out of spec, and would set it aside, and flag for assistance to deal with the new item.

A more advanced strong AI bot would obviously behave in a similar manner to a human, with “what the hell do I do with this?” type response. or perhaps concern for where the contaminant came from. But, for now, we’re not dealing with strong AI, just weak AI with strong elements.

If you were unlucky, it wouldn’t see the significance of something so thin and let it pass.

The thinness wouldn’t matter. It’s a differrent configuration, outside tolerance. It’s not supposed to be there. If it found a bloody human ear on the conveyor, you’d get the same “What the hell? Help!” reaction.

 
Flag Post
Originally posted by vikaTae:

It is cheaper to buy a baxter than to employ a human for the same position, and support is part of the contract. If used en-masse, it will definitely bring manufacturing back to the US, as these robots can undercut even most third world countries as far as wages are concerned, and bring prices down to rock-bottom low.

However, this means our manufacturing process will be largely automated, and if successful, even as the number of manufacturing industries in the US increases, the number of jobs in manufacturing outside of management, is likely to go down, not up.

Is such a trend a good thing for the US? Is it better to attract all the industries we can, even if more jobs will not be created, or should maximising jobs not industry be our primary focus as a nation?

Production is already very automated in most countries even in 3rd world countries. Basically Humans are only used when they can not efficiently be replaced by machines. Which is usually the case when either high flexibility or the still not replicated traits of hands and the human brain are necessary.
Most other human traits have been replicated or even surpassed. And the history of the industrial revolution is filled with examples where the replication or surpassing of such traits has lead to both an increase in production but at the same time a loss of jobs(where the machines replaced humans).
Personally i don´t see Baxter as a great revolutionary step in this process that could rival even something as simple as the introduction of the automated loom. As said most tasks that can be automated are already automated and Baxter seems rather limited in ability when it comes to replacing even more Humans or even competing against already existing machines.
The only thing “new” could be called the interface and learning ability, which is not really new except perhaps as a package.

To your question regarding if such automation would be good for the wellbeing of a nation(and especially its people) even if does not bring jobs or even cost them. The answer is yes. As the industrial revolution has shown even if jobs are lost due to automation, the increase in productivity of the jobs still in place will still increase the overall wealth, the problem(s) that arise are typically more results of bad/unbalanced distribution of said wealth.

 
Flag Post

Well, warehouse cobots have been around for a while, and are quite commonly used now, so warehouse staff can be safely halved. Robotic cleaners are already available commercially, and bring the same benefits as robotic line workers – no wages, utter dedication to the job etcetera.

Delivery drivers will likely start being automated within the decade. We already have some self-driving cars on US roads. Maintenance staff will still be human, as will the facility nurse, and the upper management. But as the total staff count decreases, so does the need for cafeteria staff, and things like HR and payroll start to downsize.

It’ll be great for the economy, but long-term, blue collar jobs will increasingly disappear. It is whether or not this is going to be a problem for our society.

Shipping and distribution used to be my game. Automated warehouses, driverless fork lifts etc. are a big investment. They require the installation of a complete system all in one go. That is beyond the means of many smaller companies, although like anything else, when the uptake increases the price will drop. But there will always be situations where robots of any description are not suitable. To eliminate drivers from vehicles will require a huge government investment in roads, apart from the costs to industry. It will be a long time coming, and again there will be circumstances where they are unsuitable.

Mmm, so would government re-education programs perhaps be the best way forwards in dealing with such increasing robotisation? Find jobs for these people in other sectors,such as retail where human-human interaction is still absolutely vital, or clerical work which is difficult to automate?

It’s not so much that blue collar workers will disappear but unskilled jobs. It’s already been happening for years. Skilled workers are thin on the ground, and there will always be a demand for them. Government re-education, as in re-educating the government, would be a good idea. In recent years, there has been far too much emphasis placed on giving students a mickey mouse degree which impresses nobody. They are finally waking up to the fact that we need to concentrate more on high-tech manufacturing. Training schemes are available, but it’s all a bit half-hearted at the moment, and the current economic crisis is doing nothing to improve their enthusiasm for it. But it will come. I see one of the challenges to be integrating robotic and human workers successfully. Using robots for cost reasons alone will lead to disaster, but that will be a temptation to many organisations, especially those which are led by accountants.

The story of the industrial revolution was largely one of social upheaval brought about by change, and it’s going to happen again. So I think it is very important that the both the government and industry get on top of this problem before it gets out of hand. If the process can be managed in a controlled way, the results should be very positive. It will likely alter the entire social landscape, but that’s what humans have always done.

 
Flag Post

Interesting idea, but if it becomes cheap and practical enough, isn’t it very possible that it would just replace the jobs that would have been filled by U.S. citizens? It definitely has its advantages, but the idea that it would begin to replace people could have its problems.

Besides, in the near future, we could have robots building robots!

 
Flag Post

I still think that in the end most jobs will disappear and the only way I see to deal with that is by increasing taxes and welfare (you are paying for the rights to trade on the market and the resources you use ext.). Another short term solution is probably going to be to deduce working hours so that multiple people can share a job.

 
Flag Post

I still think that in the end most jobs will disappear …

That’s what the Luddites thought, and they were wrong. Increased mechanisation in the nineteenth century created far more jobs than it destroyed. However, I do see that with robots things could turn out differently. If they become too good at replacing humans, we could reach a tipping point where jobs do start to evaporate permanently. On the other hand, if humans start to become bio/mechanical/electronic hybrids, robots may cease to be relevant in a lot of contexts.

If unemployment figures start to rocket, it would be a relatively simple matter for a government to pass legislation outlawing robots in certain workplaces. That would probably be very popular among the workers too, and governments always like to be popular. It may sound far fetched now, but as I said, there is going to be a period of great social change when new ideas will take root. New problems will need new and sometimes radical solutions. None of us can foresee what the future holds, and it will probably be different from anything we can imagine now.

 
Flag Post

Coming back to this thread was long overdue. There have been some great points here, and bauval’s last post was very much in line with what I was wondering when I made the OP. It’s the kind of input I was really seeking. There are a few other issues in the posts that I’d like to query, too.

JohnnyBeGood

Personally i don´t see Baxter as a great revolutionary step in this process that could rival even something as simple as the introduction of the automated loom.

True. What I see as the leap this particular robot has taken, is in it’s human interaction qualities. Safe to be around, safe to interact with, able to program it by gesture and most importantly, its capability for self-learning. It is only a start, but it is a very interesting start for those reasons.

As said most tasks that can be automated are already automated

I would disagree with this statement. Any task that a human can do can in principle be automated. If we can make our AIs smarter and more and more flexible, this is what increases their utility – and why I see this type of interaction medium as a great step forwards.

To your question regarding if such automation would be good for the wellbeing of a nation(and especially its people) even if does not bring jobs or even cost them. The answer is yes. As the industrial revolution has shown even if jobs are lost due to automation, the increase in productivity of the jobs still in place will still increase the overall wealth, the problem(s) that arise are typically more results of bad/unbalanced distribution of said wealth.

Thanks, that is the sort of answer I was looking for, and gives me a lot to think on, even if it does not give a lot to reply to.


Beauval

driverless fork lifts etc. are a big investment. They require the installation of a complete system all in one go.

Perhaps. The cobot system KIVA certainly works this way. However other systems such as self-driving trucks, do not. As you may or may not know, things have advanced to the point in self-driving vehicles where some US laws are backing driverless vehicles. It is not going to be much different than purchasing any other van, just that it would obviously require one or more unloading bots onboard and a crane arm of one sort another. Alternatively for primary level from manufacturer to distributer, all that would be at the depot.

The on-truck system does not require the installation of a complete system in one go, but does completely remove the need for a human driver for all but end-user deliveries.

To eliminate drivers from vehicles will require a huge government investment in roads, apart from the costs to industry.

Not at all. They were originally developed for use in combat situations, where the roads may well be blasted to hell and back, and their condition was certainly not guaranteed. Saves risking a human driver. On our roads, they’re fine just the way they are, no additional investment needed.

So long as the vehicle can get through the space, it is sufficient, same as for any human driver.

Government re-education, as in re-educating the government, would be a good idea. In recent years, there has been far too much emphasis placed on giving students a mickey mouse degree which impresses nobody. They are finally waking up to the fact that we need to concentrate more on high-tech manufacturing.

Yes, government foresight tends to be nonexistent. It is this transition phase between the disruptive wave hitting, and the government starting to take action which worries me. That’s why I’m predicting large amounts of social unrest as the technologies far outpace sluggish and uncooperative, often disbelieving lawmakers.

The story of the industrial revolution was largely one of social upheaval brought about by change, and it’s going to happen again.

Yes, there’s no doubt about that. There are too many disruptive technologies all maturing at once, just like last time. The difference here is there are far more of them than during the industrial revolution, and they are coming much closer together, which strongly suggests the social upheaval is going to be more abrupt than last time. Much more so.

 
Flag Post
Originally posted by beauval:

I still think that in the end most jobs will disappear …

That’s what the Luddites thought, and they were wrong. Increased mechanisation in the nineteenth century created far more jobs than it destroyed. However, I do see that with robots things could turn out differently. If they become too good at replacing humans, we could reach a tipping point where jobs do start to evaporate permanently. On the other hand, if humans start to become bio/mechanical/electronic hybrids, robots may cease to be relevant in a lot of contexts.

If unemployment figures start to rocket, it would be a relatively simple matter for a government to pass legislation outlawing robots in certain workplaces. That would probably be very popular among the workers too, and governments always like to be popular. It may sound far fetched now, but as I said, there is going to be a period of great social change when new ideas will take root. New problems will need new and sometimes radical solutions. None of us can foresee what the future holds, and it will probably be different from anything we can imagine now.

I see it as a big difference though. Back then, the mechanization increased what a person could do when they themselves did the work, now the mechanization isn’t about increasing a person’s productivity, it’s about replacing the people who are doing the work.

 
Flag Post

Perhaps. The cobot system KIVA certainly works this way. However other systems such as self-driving trucks, do not. As you may or may not know, things have advanced to the point in self-driving vehicles where some US laws are backing driverless vehicles. It is not going to be much different than purchasing any other van, just that it would obviously require one or more unloading bots onboard and a crane arm of one sort another. Alternatively for primary level from manufacturer to distributer, all that would be at the depot.

The on-truck system does not require the installation of a complete system in one go, but does completely remove the need for a human driver for all but end-user deliveries.

The KIVA system is the kind of system I have encountered. It works well in situations where it can deal with uniform load sizes, uniform pallets etc. Bar coding helps too. When you have to receive a complete factory piecemeal from maybe 200 different suppliers, check everything, repack a lot of stuff and make sure it is properly marked, deal with ever changing customer requirements and then ship out all the component parts in the right order, it’s a non-starter. Robots are nowhere near sophisticated enough to do that yet.

It’s the same with delivery systems. Robots are like computers, they are good at doing repetitive tasks. Depot to depot traffic with uniform trailer loads is easy. Dealing with a constant stream of goods, all of different shapes and sizes, is not something I believe thay could handle efficiently. No doubt that will come.

As for driverless vehicles, they will require social acceptance as well as technology. We have had driverless trains on the London tube since the late sixties, but passengers were unhappy with the concept. So a driver had to be put in the cab, even though he wasn’t needed. That still goes today, although more recently it has become a union thing. Terrified of losing union membership, men like this will always stand in the way of progress. Even on the fully automated Docklands Light Railway, which was heavily used during the recent Olympics, there has to be a human driver somewhere on board “just in case”.

I see it as a big difference though. Back then, the mechanization increased what a person could do when they themselves did the work, now the mechanization isn’t about increasing a person’s productivity, it’s about replacing the people who are doing the work.

Sometimes true, sometimes not. Inventions like the steam shovel were able to do the work of dozens of men armed with shovels and wheelbarrows. On the other hand, processes like drop forging still required human operators, but speeded up production. And you’re right, a lot of menial tasks are going to be given to machines in the coming years. We have to concentrate on retraining people to do more complicated jobs. We have a severe shortage of engineering graduates at the moment, and this problem needs to be addressed urgently if industry is not to stagnate. It will need government intervention in secondary education, and I don’t see too many signs of that happening right now.

edit. ElBandito, you may find this interesting