Community Combined Robot Build
Hey everyone - there's been chatter about potentially working together to build a robot as a combined community effort. I don't believe anything like that has ever been attempted (or accomplished) before, anywhere. It would sure make waves in the industry! The idea was inspired by conversation about "struggling" or "failed" robot companies. Everyone has opinion feedback on what they believe the cause of the industry stagnancy is - and there's no right or wrong opinion because at this stage of the industry, different people/companies/products are affected by different industry challenges.
And, as a optimist, I suggested "Hey, there's no shortage of industry complaints and challenges, so why don't we talk about how to turn it around and build a robot that proves us right". If someone else can do it, we can do - so I'd rather it be us
So, like, wow... I get pretty stoked thinking about this whole thing. If we come up with an agreement on what a robot should do, we can split up the responsibilities and combine them into one project. It'll also help the software grow because I imagine there would be some interesting requirements.
On that note, I've moved all of the chatter on this topic here. What do you think? Let's work together and make something awesome!
Right now, here in the US, most of our Stop&Shop grocery stores use a robot that goes up and down each isle, checks for the products on the shelf, alerts someone if a spill has happened, and will even look out for shoplifters. I think he is known as Marty the robot
Have any of you considered making robot products that someone buys (like start a company)? One of the alternatives to the billing system was to close ARC down and make a robot product, but that doesn't align with my grand vision. I'd prefer to support any of you to make robot products :). You know, we'll work with fantastic white label pricing on large volumes
PS, Nink - there's soooooooo many users that don't post on the forum who have already subscribed. I'm actually blown away at the positive response. There's a slight confusion regarding the incentive to monetization - meaning, our passion is to continue investing back into the software to make the World's best robot software better. It's not about benefiting ourselves with greed and buying new cars. I've been financing Synthiam (and ez-robot prior) for almost a decade out of my own pocket. Sure, we've had investors but I've been the primary investor and still am. It's not sustainable to continue giving software for free any longer. Any amount of revenue is incentive for us to continue truck'n on. It demonstrates that you all value our efforts. It demonstrates that you support our initiative and it adds value to your life. If no one purchased the software, I'd have to shut it down in a week or two. But, the positive response of subscribers lets me know this is something worth investing more into.
So the short summary is we're going to throw more resources at this and make the software even better so your robots get even better.
As for PoC hardware - we're investing effort into partnering with more hardware vendors to provide our customers with a larger selection of hardware. Such as Rock Pi X (which is my new favorite). We're continuing to design hardware reference designs as well - and anyone can manufacture them.
Eight years ago when I started designing Alan, I thought consumer robotics would be huge by now. When I did Kickstarter in 2015, robots like Jibo and Buddy were gaining interest, but by 2018 Jibo was a massive disaster and I could see the gap begin to form. In 2019 when I attended CES, I spoke to hundreds of business people and investors and the mood was "wait and see". But over all, the robot mood was gloomy.
B to B is the only growing sector (thanks in part to Covid) and I find that sector to be run by companies with deep pockets that develop specific task oriented robots in house. But, even that sector is running into problems when robots need to interact with humans or are just too slow to compete with human workers in the field. Elon Musk had a Tesla factory shut down to install hundreds of robots, only to shut it down and rip them out because there were so many bottlenecks in the production line.
After spending literally 100s of thousands of dollars of my own money on robotics over the last decade, I can say that I am disappointed in the flat line trajectory of robots. I believe there is some missing link to making robots accessible, inviting and affordable to consumers. The prices for raw parts to manufacture robots are slowly coming down but sticker prices on complete robots like Spot are still way out of reach for the average consumer.
Dj has said he would never do hardware again. There is a reason. Its expensive to keep the factories going, the investors fed and happy. And what happens after you saturate the market? Robots are just hard to do successfully in the current market. Education, toys and robot sweepers is all we really have to show for the past decade in consumer robots.
Take away? DJ is smart sticking to software. More so, that he has monetized it. Coming up with a prototype robot for a task, does not guarantee that an investor/s will think its a "good idea" and invest. So after months of work and lots of personal financial investment, you are deeper in debt without anyway to recoup your investment. Personally, I am slowly backing away from the dream that robots will be affordable and in every home anytime soon.
I will continue to build robots that interest me but not necessarily a prototype for a manufacturable product. I wish luck to all those that attempt to do so. I've done alot of hard work and research over the past five years and I can speak from experience, it is expensive and very difficult to do so successfully.
I like to watch that show Shark Tank because of the entrepreneurial spirit of the show. The first question the Sharks as is "what problem are you solving" in order to make it a business. The future of consumer robotics is centered on what personal problems they can solve. A friend to talk to, I have humans for that so no need for anyone to pay money for a robotic friend that I develop.
Oh - that gives me an idea. What if we all worked together one one robot? Or many... but like, if we all agreed on a specific problem to solve, then we could divide the robot into smaller components. And we can solve those problems individually and share our experiences on here?
So I came up with that much - anyone want to kick off a problem to solve? Don't think about the robot (yet). Just think of a problem that a lot of people have and then we can figure out how to solve it.
Laundry is a big one! The whole system of laundry needs to be dumped on its basket. My laundry folding robot and other laundry folding robots are novel but there's a real issue of time that needs to be solved there. From where to place dirty laundry, sorting lights/darks & delicates/non-delicates, to moving from a washing machine to a drying machine (could they be the same machine?), to folding or hanging, and back to storing in a closet or drawers.
Dishes would be my next request for problems to solve but laundry would be my first choice. Car maintenance would be another one.....you guys gotta stop me before I get too crazy lol.
others;
-mow my lawn
-rake my leafs
-shovel my snow
-clean the kitchen
-wash my clothes
-take the garbage out
-clean up clutter off the floor that a Rhoomba won't
-clean the fish tank
I’ll continue to support your software requirements to make robot products. And we can be responsible for the robot revolution.
If someone else will be able to do it, why can’t we? I’m made of the same material - blood sweat and beers
I like The software here, its been fun to use, sometimes fustrating, Im no programmer, I rather build the hardware, thats what I do best. I always need help with programming. One part of the programming I totally lack, is in how to give the robot its sorta own personality, Im more just utility minded, cant invision my designs being like an C3PO or B9. I need help in those areas.
The Invention of the EZB hs been the greastest thing ever I think
Commercial robots have a much higher probability of success especially when they are built in conjunction with a company who actually has a business need and there is a return on investment. It needs to be cheaper to perform the task with robotics than doing the task manually.
Walk into your local McDonalds and someone is paid to open a bag of French fries, tip them in a basket, put the basket in oil, set a timer, wait 2 minutes, take the basket out, tip it into a tray, add salt, mix them up, put them in a cardboard container and put the container in a bag. You could easily build a prototype robot to do this for around 50K. If you built robot fry machines in bulk maybe 10K each. If it is a busy 24*7 McDonalds the cost for Labour to do that is probably over 100K per year. No sick days and an easy return on investment.
There is about 20 bulldozers sitting idle at the end of my street. They are building a new subdivision. They work Monday to Friday 9am to 5pm and today is Sunday so no work is being done. For around 100K we could add robotics and remote control a Bulldozer with something like exosphere, add some self driving automation, an AI program with the subdivision plans loaded and you could do the job in half the time with half the number of bulldozers. These large bulldozers cost approx $175K each and in high volume production we could probably automate a bulldozer for <20K each. There is an easy ROI. Even if you just remote controlled them from India using exosphere and low cost resources there is a business case. You could then use the data in production with machine learning you can teach the bulldozers to operate on their own, further improving the business case.
If we want robots to be successful we need to develop robotic solutions that automate manual tasks where there is an ROI. Synthiam is a great product to do that with as you can build a prototype very quickly.
1) Mapping and autonomously navigating the home (or for the outdoor options, yard).
2) Returning to a home base to recharge the battery.
A few years ago, object recognition would also have been on this list, but I think we have made huge advancements there in the past couple of years and are close enough that at least some of these uses are possible.
The one I would add, that would take significant advance in object recognition would be "weed my garden". A task that I detest doing, but even I pull out the good plants half the time and not just the weeds, so a robot would need to have excellent recognition (If Google had an API to Google Lens it might work. It is pretty good at identifying what a plant is. but we would still need to tell it if the plant was desired or a weed).
I would also like to see us do more with assisting the sick, elderly, or disabled. So many seem to focus on using the robot for companionship. Jibo was great at that, but failed because it had no ability to do anything physically useful (IMHO). My elderly and physically struggling mother has no problems there since discovering Zoom and Facebook Messenger so she can communicate with friends and family and participate in events. Her issue is picking up things she has dropped, finding things she has misplaced (like her glasses that went missing for a week) or retrieving her 4x daily medications from the kitchen counter and reminding her to take them. This would not need total autonomy so much as accurate object avoidance. She could direct it by voice or possibly joystick to go to the right place, and to get back to a charger, although a roll on charger would be critical because she no longer has the hand strength to plug and unplug from the wall, nor the dexterity to deal with something like a DEANS connector.
Alan
I think some years ago, there was a company down south that started to (Gekko Robotics I think). they started to make a personal assistant model and a wheel chair with built in robot arm. My mom, being a stroke victim is wheel chair bound and cant speak really any more, so she now gets state provided live-in care.
Its what prompt me to build HEMI. I see a need for a mechanical device to help an aid/nurse with home care. I know they already have many devices out there but to me, some are nothing short of medieval torture devices. A robot with an ability interact with the aide and the client would be a nice next step, even adding some teaching and security features into the mix to help.
This is across the highway about a 5 min walk from my house on farmer browns corn field so it doesn't bother me as I am out of ear shot. There is also a bunch around the corner where we used to have a golf course. I think it is time to move further out of town. Perhaps self driving robotic bulldozers is a bit optomistic but there is certainly a bunch of industry robot use cases we could come up with.
Most of their requirement is entirely right out of the box and ARC + exosphere takes care of that. They have only one requirement which is a bit unique that I can’t explain due to their product privacy. But I can foresee it being very useful for their application.
we’ll make a demo video of it all when it’s up and running.
But, it’s definitely inspired me to want to build a robot with our community group as a collaborative effort. I’ll noodle some ideas around how to make it work - and what similar hardware we have to work with
We are seeing an increase in delivery robots as a result of the pandemic. They are delivering food in condo's, hotel rooms etc. Restaurants who are permitted to stay open also want to protect staff. So delivering food to tables is also a requirement.
The idea is to start a company making lightweight low cost delivery robots. People are bringing delivery robots in from China but they retail starting at $10,000 https://www.robotshop.com/en/catalogsearch/result/?q=delivery&order=relevance&dir=desc
Clearpath was going to use the roomba to build a robot but they abandoned the project and opensourced everything 3 years ago. (BSD Licence) https://github.com/turtlebot-euclid/Turtlebot_Euclid. Not sure why they abandoned the project. I assume because they really didn't have a use case and the cost to buy Roomba back then were quite pricey. It was done with ROS but could easily be done in ARC, and I heard some rumours of ROS and ARC working together in a future release :-)
As others have pointed out and are working with the create2 (roomba 600) for $199 at the moment. https://store.irobot.com/default/create-programmable-programmable-robot-irobot-create-2/RC65099.html add a charging base that's $40 . if you developed a low cost shelving unit that housed all of the electronics, camera, battery, wifi etc and a couple of shelves for delivery. You could probably build the delivery robots for < $1000. Maybe you could sell for $2000-$3000.
We could probably make the base and shelves using a laser cutter and acrylic (3D printing is a little expensive and slow) and when you get some sales make in bulk aluminium like the Clearpath robot.
https://github.com/ibaranov-cp/Turtlebot_Euclid/blob/master/turtle-bot-euclid-6.jpg
https://www.robotshop.com/media/catalog/product/cache/image/1350x/9df78eab33525d08d6e5fb8d27136e95/k/e/keenon-commercial-delivery-robot-t6_1.jpg
The hospital in my town uses robots to deliver meds, blood,and and food to all rooms and on all floors, guded by wireless sensors in the celing on each floor.
Robot dev cost are def a big factor. Ive been working on HEMI a lilttle over a year now and its cost me close to $10K US already.
Should be interestingto see what we could come up with here
I agree on the dev costs being a major drain and building a self charging robot is a huge expense but bolting on components on an existing robot should take costs and time down.
VSLAM is maturing and GPU/CPU integrated solutions like Jetson Xavier NX are coming down in price $400 and will open up robot vision to low cost robots. https://www.seeedstudio.com/NVIDIA-Jetson-Xavier-NX-Developer-Kit-p-4573.html
@DJ We we really need a containerized .net core version of ARC on linux so we can take advantage of these types of technology.
You can use exosphere to map out the environment and train and use exosphere when issues occur.
.The height is always an issue depending what floor or terrain it needs to travel on, I was able to fit all my electronics on the Roomba cover plate but looks like a bird nest all compacted on small area. so for sure will look at different body types I can interchange with...Halloween can be a Zombie or Terminator,take to hospital and could be that Futuristic body, I do believe Telepresence(did i spell that correct?) will be part of the futeure as in Japan they are all over the place in Airports and stuff as Information booth robots and like I said what sick kid in a hospital does not love a cute robot coming to visit?
Great news re charging station included. Already saved $40 we could use that for an LED display.
I don't think it is actually a Roomba under the Keenon I just thought it was a good design of what it could look like although you couldn't have that many shelves as I don't believe it could handle the weight (maybe 2 shelves). I like your idea of interchangeable clip on bodies depending on what it needs to do.
Business model could be something like $100 a week for Robot as a Service (Maybe min 3 months so we recoup costs) this way we can have people monitoring the robots, cover maintenance etc. If the use case doesn't work that's ok, they send the robot back and we use it for another use case/customer.
Ya I just found a great body for 20 bucks at Walmart to get started extremely light weight black hard plastic fits perfectly over Roomba and if too tall it can be reduced to half size as it is 2 pieces.My other Cylon project head is also just right size. Can't wait to see the Cylon vacuum and then return to charging dock shut itself down to recharge.This is a great winter project. The Ezb , battery and electronics all fit in
----The second biggest battery is 12 volt 18 Amps there ,I tried it in the body bucket,and the Roomba was able to carry on as if no extra weight was added! Half the size of a regular car battery and about 8 pounds of weight there.
we get some old jazzy wheelchairs as a base you can pick them up pretty cheap on craigs list etc. Strip them down and add a box for food a camera, cell phone and a realsense then use exosphere to control them. We can drive them remote and map out their paths and then when we have enough data let them go autonomous. I think a prototype could be built for under $1000 and we do a test run with a local restaurant .
nink. The robots that drive outside have some challenges - kidnapping and battery power. Navigating the sidewalks with a well mapped area is doable. People would just need to not mess with it. How do we solve he kidnapping and energy problem?
Theft
Theft is always going to be a problem and we can throw cameras and alarms at the problem (remember the security robot made from a jazzy chair https://synthiam.com/Community/Robots/Security-Cam-Robot-This-Droid-Is-Watching-For-You-Jazzy-Power-193 ) I guess we could disguise the GPS and with a separate data feed like Lora so at least we can find it.
If you go to a lot of cities there are hundreds of electric scooters just laying around and no one steals them. Norway, Sweden etc even Toronto has a pilot program. Even high crime areas like Sao Palo, Brazil they are everywhere.
If we started off in low crime wealthy neighbourhoods (the people who can afford uber eats) I believe the problem of vandalism or theft would be lower. Once they are pervasive people would get used to seeing them and begin to ignore them I believe.
Autonomous
If we started with a location where testing autonomous vehicles is legal we could start to do test runs on known straights or low risk sections of the journey. One person could control 2 or 3 at once this way.
Other Revenue
Once we have these on the road we can use them for a whole lot of other purposes as well. Courier deliveries, cover them with Advertising, you could probably even charge people to get to drive one (cutting labour costs).
A faraday cage (i.e. tin foil coated emergency blanket) work well for blocking wifi/gps/all radio signal for kidnapping robots. Okay, you caught me that I think about that stuff. Mostly when I see news about delivery drones or robots.. I think to myself, what a wonderful world - how could i steal that robot?
Battery power in cold weather is gonna be a biggy - so maybe Canada isn't the best place for pilot HA
Lastly, if the robot was designed to have short range, it really could operate downtown in trendy spots to serve other downtown residents (or some sort of down town area). Meaning, it would be a real neat marketing ploy for restaurants. Think about it... chill'n with your friends and ordering some pizza and it gets delivered by a robot?! Heck yes! Best party ever!
But... even with a robot delivering the food, you're in competition with monster well-known companies like skip the dishes, uber eats and door dash (apparently they're still a thing). Guess you'd have to work with one of those companies to get on the radar of people?
Okay so logistics aside - maybe that's something solved with pilots and such. Is this the best robot? Better than an affordable robot that cleans floors in malls/supermarkets/etc? Or a door greeter telepresence at offices that asks covid questions and brings people to their meeting autonomously?
The problem with robots in malls etc is they don’t want autonomous robots with consumers around. The greeter robot sounds great but when a customer is hit by a robot and sues the corporation they could be liable for millions. This is why we don’t see them running up and down the isles in Loblaw and Walmart even though they have been testing them for years.
Cold weather can be solved with heater batteries, Tesla is a good example. They drive them in Canada in the winter (why on earth you would drive a low profile Tesla in snow though is mind boggling)
yeah you would partner with Uber eats etc they would rather pay $5 for delivery than $10.