The Evolution of Stretch | Boston Dynamics
Вставка
- Опубліковано 7 лис 2023
- Stretch has moved over a million customer boxes in under a year, improving predictability and preventing injuries. But how did we get there?
Discover how we put our expertise in robotics research to use designing, testing, and deploying a warehouse robot. Starting from the technological building blocks of Atlas, Stretch has the mobility, power, and intelligence to automate the industry’s toughest challenges. - Наука та технологія
![Evolution of Boston Dynamic’s Robots [1992-2023]](http://i.ytimg.com/vi/Rdm2ggtFvmQ/mqdefault.jpg)
![Evolution of Boston Dynamic’s Robots [1992-2023]](/img/tr.png)
Nice to see the evolution, and both Stretch and Spot finding real applications.
I hope also Atlas finds its place soon..
I agree, Clive. It’s definitely interesting to see how Stretch became so uninteresting in the end lol. I mean that in the best way, though. It went from a dynamic, spazzing ostrich to a rolling box that looks like it belongs in an Amazon warehouse - exactly as it should.
Hoping to see the day robotic technology get so common that can reach the test bench of Clive to be reverse engineered.
Handle is a very cool concept which was too easy to be a solution looking for a problem to solve.
Luckily the engineers recognised that they needed to look at the problem first, take their baby behind the shed and start over with an actual solution.
Stretch will find it's real world application, because it was actually designed for it. (Something Elon keeps forgetting to do)
Hmmm... i find them rather shallow and pedantic.
3:32 - Although they ended up not being the right solution for the warehouse environment, those ostrich robots are my favorite design. The motion they exhibit in order to maintain balance is one of the most natural motions I've ever seen in a robot and makes them look almost unbelievably alive.
It would be awsome if they turned it into some sort of smart offroad vehicle with the capabilities to pick up objects like logs or whatever on your adventure(the first model with the hands)
Robots twerking to do their warehouse job. That's the future we could have had if BD had stuck with it. 😢
@@echthroi9 It looks inefficient and unreliable. And expensive. The flat base of Stretch is better where you have level floors.
It's just not elegantly moving. I imagine if it was taught martial arts, it would probably have some pretty good kung fu and parkour skills like the wheelie people in return to Oz
@@hyperverbal w in the h f are you talking about.
If you could make a Stretch that could handle suitcases/luggage for airports - you would make a lot of airports happy.
People hate AI because it replaces jobs, but thats lowkey our only option to compete with the likes of China or India.
All the jobs that could be automated, have been outsourced anyways. We have nothing to lose, and everything to gain.
I firmly think we need some sort of UBI in the future though once AI really takes off.
I hate socialism so maybe instead we should have UBI for "fun" jobs so people wont get paid for sitting around.
Psychologically being unemployed without hunting/gathering might be bad.
I think jobs like park rangers, arts, Research, and community stuff, should be subsidized and created as a UBI imo.
@@honkhonk8009 To quote a famous mad scientist "Where we're going, we don't need jobs"
@@honkhonk8009 tech would be stolen by China though.
@@honkhonk8009the problem is more that the idea of needing to work a 9-5 in order to survive is a very outdated concept in an era where one person’s work can provide for hundreds.
@@honkhonk8009 That is actually not a bad idea for a futuristic UBI implementation.
As someone who has worked in warehousing, stretch excites me more than any of the humanoid robots.
feel ya
@@user-yv4gg7jb2f As someone who works within the escort service industry, stretch excites me more than any of the humanoid robots.
hopefully we can keep our jobs for a few more years.....
They terk err jerbs!
@@Chris-Phantomview 5 years?
Stretch is great and I love stretch, but still I hope one day Boston Dynamics can continue to evolve Handle. I love the wheeled motion!
The ostrich was the coolest of the designs.
There are a number of applications I’m curious about:
1. Narrow very big boxes like servers- not much room to grasp from the side, heavy, delicate.
2. Boxes with shifting weight like cat food or other particulate item
3. Damaged heavy boxes: torn corners, tape coming undone…
4. Wet boxes
5. Reaction time: unloading a truck and someone surprise comes around the corner while a box is in motion. (improbable situation with safety standards, but still)
6. What happens when it’s not the robot’s fault a box breaks: soggy bottom and product falls out: protocol? Damage prevention?
7. Does the charge vary based on load? If it is relatively fixed in position can it be directly plugged in for no charge breaks?
8. Oil/grease/maintenance
I would guess that lubrication/maintenance is not more than other mechanical stuff in a warehouse, conveyor belts, forklifts, etc.
@@bennylloyd-willner9667 I think that’s a reasonable assumption
leaving the comment for be subscribed to the main comment's answers
Interesting questions. I don't know the details.
If I made a guess. The machine learning could be trained to detect this stuff as well and avoid it or raise a "help me" message and shut off. If it detects a box collapse. Just give it some bad boxes in the lab record the data of the breakup in a controlled environment. then use that data to train the AI on these situations.
Damaged boxes were mentioned. Working retail, I’ve seen some boxes on pallets that really seem to have had the snot kicked out of them. Especially if something leaks on one of them and the cardboard starts deteriorating. Many things come in cardboard trays with plastic wrap on the top and if that plastic starts ripping it becomes unsound to support the item. This may not be a deal breaker if it can recognize anomalies like that and summon someone for rare items. The vacuum gripper could probably handle the vast majority of things but sometimes one hand just isn’t enough.
It seems like a well designed critter. I hope it works out.
Yeah, any particularly heavy or compromised boxes picked up from the top like that seem likely to just dump the contents out of the bottom or have the top rip off.
I agree to an extent. I will say I worked in the storage in the back of walmart for a single day (I was a cashier that wanted to try out other better paying positions and they allowed me to try it and I didn't like it at all.) And people were dropping boxes, throwing them to each other, and just not seeming to care. This is another instance where if the robot damages 3% of boxes badly enough to damage the content of the box, but humans damage 5% of the boxes that bad, then this is still better than a human, as humans are not perfect and make mistakes so the robot just needs to make less mistakes.
@@GamingWithNikolasi am going to hazard a guess, you FIRST concluded the robots were better than the humans 😉
How's that working for self driving cars in San Francisco 🤭🤐
@xaxfixho I was just pointing out that robots don't need to litteraly be perfect in every situation to be usable. It just has to do better than humans.
Assuming the plastic wrap has been slashed, lets say for a pack of pepsi cans. Well wouldn't the vacuum gripper not do an excellent job of holding the plastic wrap together still? Basically the weakness that is caused by the slash to the plastic is redundant because the plastic is being held by many grippers all over. Basically many "hands" holding the plastic together in it's initial position.
But if the plastic has actually fully opened, then yeah, then you're boned... If we can get to this level of automation though, maybe it just makes sense to have a "spillage channel" and the machine just moves on to the next package, allowing the unpacked product to fall off into a waste disposal?
I've worked in logistics and distribution for 20+ years and I've unloaded many shipping containers full of thousands of boxes. I can relate to the fact it is not fun unloading containers in 100+ degrees. It's really exciting to see what you guys are doing. I'm sure we're a few years away from these robots becoming mainstream, i.e. the cost of these versus cheap labour, but this is a really important innovation. It's also cool to see BD robots being put to practical use, as with the Spot videos, much as we all love the dancing/gymnastics Atlas videos too 🙂
Do you think they are going to viable long-term 🤔
They are competing with cheap labor that can adapt to different tasks.
How much do you think it's costs to buy one vs hire some cheap labor.
Returns on investment?
Yes I do because eventually these will speed up to point they're doing the work of 5-10 people, accurately. They don't need HR/payroll/managing/security and they do it without complaining. I lost count the number of times we had guys call in sick when they knew multiple containers were coming in that day. These machines don't get sick (assuming no regular breakdowns). If they can stack pallets rather than just throw on a conveyor, which I've seen them do in other demonstrations, then that increases the return on investment even further. And labour in the UK is not that cheap even when it's cheap.@@xaxfixho
@@xaxfixho cheap labor can adapt to different tasks, but cheap labor cant work 24/7 year round. I can only assume that eventually the cost between a robot and human labor would equal out or eventually be better for the robot, because the robot is able to make up difference through infinite work hours and durability in weather conditions (extreme heat or cold). And a robot won't complain about working conditions..
1:06 '... and make its own decisions about what to do next'
Strech: 'aight imma head out'
Great job, and great to see practical applications in warehouses! I imagine getting a Boston Dynamics robot to do maintenance work such as plumbing will be the most challenging job of all the types of robot work, but I do really look forward to seeing that level of achievement.
All I heard was *_"Robot"_* and *_"Suction"_* .. the future is looking bright!
Profile pic checks out
@@measlyfurball37 Finally .. somebody gets me!
Ive been watching this channel since the bigdog update in 2010 when I was a little kid and my love for these robots have only grown. As a now college student this just makes me want to work harder because Id love to work with robots and possibly one day Boston Dynamics. Ive been amazed at the grown of these robots and how people have worked so hard fo improve these and how fast these have improved. I adore these robots and Id love to learn more and continue to see where these guys go!
Sometimes the best solution it's the simplest. Interesting to see how they went from humanoid robot all the way back to wheels+arm.
Imagine if a calculator was shaped like a brain 🧠 😳 💀 😅
Or a car was shaped like a horse 🐎 😂
Or a boat was shaped like a fish 🐟
Big fan of BD. I'm sure Stretch is causing a lot of excitement in the logistics/shipping world. I wonder if the speed/efficiency could be enhanced by having two arms. While one is putting a box on the conveyor, another could be picking up another box. Seems like there is enough room on the base, just would have to have the compute be able to choreograph the dual movements in the confined space.
It's called buying 2 of them.......
@@johnsmithe4656I suppose they mean having two arms on one base. Having two would indeed take up a lot more space. I think a two-armed version would be worth more than two one-armed versions
@@johnsmithe4656 Can't fit two of them in a semi trailer.
If you look closely when the arm is on one side the lidar scan the other one so it always as a fresh scan for every box placement and the environement. This way if one box fall or something happen when moving it will instantly adapt the next move instead of doing a mess until the next scan ;).
So with two arm you'll need two lidar but the movement will be slower because the arm will still have to wait for the lidar to scan before taking the box so the second arm will not take the box while the other places it and also every movement for both arm will not take the same time to achieve so one arm will delay the other before alowing the platform to turn at the end it will take more time to turn and will cost more money to make.
@@Skidline_ I didn't say it would be easy to just add another arm and I do realize it would need two lidars. Why would the arm picking need to wait for it's lidar to scan. The data from the other lidar could easily be used when it picked up the previous box. The idea is to get both working symbiotically. Even though not every pick and placement will require the same time increment, (i.e. placement on the conveyor would likely always be less than the time to pick), you would be cutting the time to turn in half because when it's turning to place, it's also turning to pick up. As far as the cost, even if a two arm Stretch only meant a reduction of 1/3 versus a one arm Stretch, that still is a huge savings that will pay for itself.
Boston Dynamics is the only company am hyped to watch their ad in recent times. Nice to see the advancements going on.
Thanks for this. It’s really important to showcase real world applications that are more approachable to the layman. Even Spot has a low real world-to-demonstration ratio, at least in terms of what is showcased on this channel.
If only maintenance was also accessible to the layman lol
@@quantumblur_3145ultimate vendor lock-in 🔐
Especially vs cheap labor, NOT going to work 😂
@@weird-guy showcases exist to show its specific strengths in real-world use cases so investors know what they're getting into
Handle seems like the perfect emergency response bot for places like large malls, theme parks… etc. where a visitor might have an allergic reaction or something and need medical attention very fast. Those ostrich legs with wheels could easily and quickly maneuver through crowds, and large places to get there fast.
Even if it just has instructions for cpr, a defibrillator etc to allow the people on location to wait for help that would be huge!
There’s a ton of really important and interesting use ideas for a fast semi-humanoid robot with wheel feet that can maneuver fast.
great times ahead ! no more grueling tasks to do !
This is the type of work I wish I could've been a part of with my Robotics degree... but hey, designing substations is good too. Can't wait to see what else you guys cook up!
You are the master of your destiny
@@MegaSuperCriticdestiny and self have 50/50 custody of the next destiny
@@MegaSuperCritictell that to all the robotic companies that reject me from entry level jobs due to them wanting more work experience
@@Jacksontruman92 That's sad and a bit demotivating. Anyway, keep it, hopefully someday you'll get the job of your dreams!
@@Jacksontruman92 I feel you, some companies don't understand you can't be fresh out of university full of ideas and knowledge, and have "seven to ten years experience in a narrow industry field". It can be difficult and demotivating, but don't give up. Somewhere down the line, someone needs you - not only your skills, but also your side of the dreams and wishes.
I'm glad that Boston Dynamics is moving forward and Stretch seems to be more efficient, but, I wish I got to see more of Handle since second to Spot, that was my favorite design
the trajectory on all these form factors is insane
Nice to see Gigachad is working in Boston Dynamics
I hope robots like this will be everywhere in ten years time, replacing all the grueling work low-paid workers have to do today. Make the robots work, and let us humans lean back and enjoy the fruits of thousands of years of human innovation.
Yea sure, because they would allow us to procreate uncontrolled when human cattle is no longer needed.
Thanks for releasing this vid! So cool to see as an outsider.
Looking great BD Keep up the great work, We Love stretch and Spot and Atlas.
Keep it simple! Always a winner.
I loooove Handle, easily my favorite BD robot. Just so cool.
Fun to see the Evolution and why the humanoid idea is not allways the best idea. handle looked very cool though.
It is sorta humorous how they started with such an advanced design then optimized it into a simple robot arm on a rolling pedestal.
I suspect the humanoid shape would be more suitable in the future for a robot that can do a little bit of everything, not only moving boxes.
@@petal9547 Absolutely! Specifically, once AI is incorporated into such robots enabling complex task via simple verbal commands.
Seeing Atlas, Stretch and Spot together instantly gives me MST3K vibes; shout out to Crow, Gypsy & Tom Servo!
amazing product. beautiful design. great work BD! True innovation
Really cool. I wish I could work at Boston Dynamics.
You are making great progress :) :)
Thanks for making me dream for more than a decade and for building the world of the future
I think, working for a company like this, testing out the edges of technology, would be one of the greatest jobs anyone can have... 😎
So awesome that it became an actual product that's being used.
0:40 Don't get tired. Doesn't get injured. Doesn't Unionize. 😅
😂
Seeing the three robots next to each other... Would be funny to see a Love Death + Robots parody with them...
Many minds with a fairly open budget focusing on making the best version of a robot. Cool.
Impressive robot, impressive company!
I can't believe that vacuum system is actually effective in the wild. It seems like it would constantly be defeated by dust, misshapen boxes, heavier items -- pretty much anything. And that says nothing about the wear and tear on the "suckers". That's a boatload of duty cycles for a flexible material. Though I guess by the same token they can be made relatively cheaply and just replaced when they go bad.
It would never work in the real world. Too many variables...
@@mikeblazthere are niche markets, where all that is known.
It's like tesla truck 🚚 vs regular diesel truck. Will only make sense in very small markets. Unit cost ,plus battery replacement costs not withstanding
Could it do twice the work with 2 arms? One could unload while the other grabs more boxes, and the whole top could rotate. Love it as is too, its great to see the field of robotics advancing!
That's twice the movement to balance, but if anyone could manage another arm it'd be BD
@quantumblur_3145 True, but easy these days. The two arms would counterbalance each other to some extent, too. The near double productivity is the real benefit. Maybe for version 2.0
Two robots working together could do that
@@nathanbirks8876 like that guy in sprited away XD
@cursed_cats5710 or 2 robots in one, lol. You would save $ on the base and they wouldn't have to move around each other 😉
Employee: Stop touching my boxes!
Robot: I'm afraid I can't do that.
I think it stops if you get close to it
That's what engineers do. They print negative linework drawings of the robot they're working on. That's just what they do.
You guys are amazing 👏 👏👏
I would pay so much money to have a smaller version of this doing dishes! Give this thing a claw and it could basically be a roomba going around the house, putting dishes into the washer and unloading it when that's done. Maybe it could have a layout of where certain items go and it could clean up the house, pick up laundry. If the whole idea is loading and unloading, that's basically the chore of doing laundry minus folding stuff. An interesting challenge I hope we one day get to see a solution for!
The company you are looking for is definitely TESLA
This is really cool
Cool to see all the robot designs you had to iterate before making a commercial robot.
Stretch is Awesome!!!
A South Korean man in his 40s, who worked for a robotics company, was crushed to death by a robot after the machine apparently failed to differentiate him from the boxes of produce it was handling.
The man was inspecting the robot’s sensor operations at a distribution centre for agricultural produce in South Gyeongsang province.
The robot was lifting boxes filled with bell peppers and placing them on a pallet when it malfunctioned and identified the man as a box, according to the Yonhap news agency.
accidents will always happen, in regular warehouses accidents are quite common
Sad to hear 😢, we need more safety for the workers 🙏
@@weird-guy the sun also rises with or without robots.
I also y-e-p my rz with or without robots.
That's a non statement, changes nothing.
Add nothing to the conversation
thats probably why stretch has an auto-safety lock that goes off when your near, so it doesnt accidentally confuse you for a crate
Boston Dynamics makes the coolest products on the planet
This is by far my second favorite company ❤
Your current boss will be proud 🙌 😂
7:17 love that placeholder product
Can we get video of it unloaded a pile of boxes that have fallen over?
Nice one :-) saves people from unloading trucks
I know they can't love me back for a long time, but I still love them, I can wait.
Stretch: Preventing human back injuries one box at a time. lol
Stretch needs more love like the others
From bipedal to printer with an arm to a rolling carrier 🎉
Beautiful
I love watching these robots move; I look at them and I feel like I've been pulled into Deus Ex or Ghost in the Shell
I used to unload trucks for Target.
Boy am I glad I got that history degree now.
offset the arm to the side of the robot, attach the conveyor belt to the robot. It wont have to spin 360° for every box. Pickup box then place the box on itself, just 90° of rotation.
2023: Still waiting for my robot butler 😒
currently for the tending of machine tools we are following the path of moving from the classic fenced cells to CoBots, which however remain "tied" close to the machine: the Strech seems to me to be exactly what awaits us at the end of that path.
Handle would be perfect for longer range logistics. Moving packages around the warehouse quickly is something only belts and conveyor systems have been able to do, so using handle would free up a lot of space and lower installation costs.
I can see the robot being used to move boxes from one warehouse to another, while stretch unloads trucks. Once Atlas is further developed, you could even throw one in to manage more confined tasks and help sort out any issues on the belt before needing a human to step it.
Spot is... Well, i guess you already know what Spot can do.
In order to be able to expand the type of objects it can grab (odd-shaped parcels), a future version of the gripper could be split in 3 squares, where the center one is fixed and the side ones can pivot. And then improve on that, maybe split it in 5, 7... basically a vacuum hand.
Pretty please put an LLM AI in Stretch too. "Would you like me to lift this box for you? And this box? And that box?" I so badly need to live in a world where all machinery can chat to me lol. But seriously I am glad Handle got to his destination
You're thinking of the Tesla humanoid robot. BD robots don't have a brain, Tesla's do.
@@jamesspencer641 no it was from another BD video where Spot was using AI and was talking. It was wearing a hat and had a few different personalities. I know it's sounds like I'm trolling you but it was real lol
love your work guys, keep it up
Impressive!
Awesome! Is it just me, or does it seem like two arms on one base would speed things up a little?
I need this kind of video about Atlas and Spot too!
Great job BD.
Thanks!
An incredibly clever company with some of the brightest minds on earth creating some of the smartest robots on earth.... People have seen amazon's new warehouse robots (which I must say are nothing on stretch).... I've done hours of lorry work and having to load/unload the back of a container... It's miserable work with a pallet truck.... I'm very very impressed with Stretch (and Boston Dynamics in general).. I have paid attention to their dog style load carriers which are trialled (as just one of) purposes to carry military kit into the field to lessen footsoldiers being wounded...
At the same time job losses are on people's mind at the moment with the sped up advancement of A. I. And it is important that staff on previous roles whom are replaced by a robot role... Are atleast offered a re-purposed role within their said industry otherwise human staff are to be rightly upset.... Just hope companies will honour their staff and give them opportunity.....
There are always two sides to a coin....
There is someone out there who will choose to go into a career into robotics simply because of this shot right here 1:37
A good example of how product evolution follows the evolution of thought about the product. Also good to see robots that can do boring stuff, so that people don’t have to. Other good mundane tasks for robots could include weeding the garden and picking up rubbish.
Excellent 👍
its an interesting robot but most of the warehouse work i have personally been involved with is either pallets coming into a warehouse and cages of stock or more pallets back out of a warehouse.
I think there are multiple players in that space. I doubt it would make financial sense, look at yhier history. Probably to their 👁 eyeballs in debt with all the custom hardware they have tried over the years
I really hope Handle could make a return, perhaps with an improved movement control system?
Im studying mechanical engineering with the hopes to do my masters in robotics and have been tinkering at home with my own attmepts at various robots, but the hardest thing ive found is not building things, which although ardeuos i absolutely love so always find the drive to persevere, but an actual use for the sruff ive built, example: i built a sara arm then couldnt think of anything to do with it, i have a bodge 3d printer i tried to make using my scara and i have a robotic arm mounted on my desk that holds an ornamentel pencil. And a janky quadraped that stumbles round my living room.
I love robotics becausr its so fulfilling to make something cool but then so unrewarding when its do difficult to find problems that need a cool solution. Reality is the things we want automated are the most mundane tasks, qnd chances are if theres a cost effective and reliable solution, someones already made an extremely optimised version.
Really cool stuff.
I have been following the project since the introduction of "handle" and have written countless emails to get an offer for a stretch. Unfortunately, no answer so far. It would be a dream to integrate a robot like this into our processes.
Exactly. Boston Dynamics robots are very impressive mechanically. However, they don't have the software brain to back it up. Thus they will always be limited to very specific and regimented tasks that require no intelligence. The Tesla humanoid robot, Optimus, does have a brain based on their Full Self Driving software. That is the brand of robot that will be in your factory within two to three years. If you have not seen Optimus you should check it out!
Good job!
Дякую за цікаве відео 👍
08:23 "then you realize... it's boring" this sentence is so sincere! The main product is there and has a market, point. But that's not Boston Dynamics... Boston Dynamics is there to create, push boundaries, impress, suprise! The application will come as the demand shows itself after your video releases. But creating is your core business IMO. Once a product has a main purpose, it's time to move on to the next one. Your value is in your craziness, guys!
Now all we need is a Boston Unobtrusive-Background-Musics.
Awesome
now we need evolutions of spot and atlas
Very cool
Small companies need hands for multipurpose tasks. Remenber that the capabilities of human hand not come from only its shape but also its matter.
LOVE
Lots of future applications for the elderly.
Grate video!
Stretch seems to be really efficient AND it also looks great. Handle is kind of nightmare fuel to me haha
being able to have a similar system with forks for pallets would be nice as cases are only one side of product packing
This is much better than dancing and backflips
I love how people make fun of Boston Dynamics for building something that isn't as generalized as a humanoid robot, but which undoubtedly excels in strength and agility in tasks unfit for humans.
Невероятно. Очень классные разработки у вас Бостон Дайнемикс. Эволюция состоит в том, что грузчик в прошлом теперь становиться оператором робота или программистом по машинному обучению.
Stretch, coming to replace a warehouse worker soon!