CE Project: Robotic Arm for assisting humans

Great start. I had a telephonic talk with xHx and I'm sure we'll hear from him soon!

Guys one thing: We need a timeline for this project. Also, we'll be selecting team members to work on this project.

Over to xHx, for details.

How do you come up with such brilliant ideas?

I've always wanted to build something like this. I would love to join in. One thing though comes to mind. Are we building a replacement prosthetic arm or are we building a automatic arm not directly attached to the person that will help them?

@ Lock - Os : The initial idea that we discussed was to build a 'portable robotic arm' that can help the disabled, old and even the kids(?) pick up items from the dish and feed them. If everyone's with me, can we build on this idea? All inputs are appreciated.

Great!!! I think we should develop such a arm that will more effieceint and easy to operate than which are available currently and artificial limbs....if we set a timeline than the team members have to allot dedicated time fo the project!!

I'm in. I just wanted to clear the idea up, as a prosthetic pretty much does the same thing and is more directly controlled.

Well, to start with, we should probably focus on what these people unfortunately cannot do, as that will drive a majority of our design from the things we intend for it to do and who it does it.

For starters, this robot is probably going to be able to move some distance to complete it's task. I can't see a disabled person moving it around all the time. That kinda defeats the purpose.

Secondly, This thing is defiantly going to need image and pressure feedback sensors and some really quick learning AI to back it up. Manual control of the robot if needed as well as programing a routine could be done though a computer or though an over sized remote with large buttons for people without a computer.

As for activities this robot will do it seems like we are heading mostly into the grasp and movement area of things. A person could use a laser to point at an object on the ground and the robot could detect it and go over and fetch it and bring it back. Simple, and quite effective but it won't be able to move food to your mouth.

Speaking of the idea of feeding people. If a person can't bring food to their mouths, I wouldn't be so sure if they could cook it in the first place, so maybe a temperature sensor and a cooking program of some kind might be helpful in preparing the food.

How bout putting sticker labels with a certain colour on objects to be picked up by the robot arm, for eg having a blue sticker on a bottle of water, white sticker on a medicine bottle, black sticker for TV remote.. the robotic arm can have colour sensors thus capable of differentiating between a water bottle, medicine bottle or TV remote..

JazlanHow bout putting sticker labels with a certain colour on objects to be picked up by the robot arm, for eg having a blue sticker on a bottle of water, white sticker on a medicine bottle, black sticker for TV remote.. the robotic arm can have colour sensors thus capable of differentiating between a water bottle, medicine bottle or TV remote..

Welcome to CE, Jazlan!

Yes, sticker labels would really help the image processing aspect. Other than colour code, barcodes can help too. Not the ones that you see in products, rather, those specialized 2D or circular codes that are optimized for image processing purposes.

Ok, I have to mention this before we move on. We want to make this project simple at first, so that the objectives can be reached quickly and practically. We do not want this project too be complex lest it ends up perpetually in discussions (like the three part marathon of CE Bot).

So, Biggie suggests a clear goal of specifically picking up food and feeding people. Lock O's has a good point about how to prepare it in the first place.
Perhaps we can assume that the food is ready before hand? Lets say someone has prepared it before hand, and food is placed before them like in hospitals. Is that reasonable?

Or, maybe we can generalize and say "any object" that the person wishes to pick up. We can use image processing.. whether through visual aids such as stickers or following a laser pointer like Lock O's suggestion.

Regardless of the objectives, we also want to know how to control it. Autonomous? Remote? Joystick? Voice? Keep in mind that we need a balance between simplicity and practicalibility.

I'm not quite so sure about the feeding idea myself. If they are in a Hospital, the nurses could help the person eat, and make sure they don't choke on it or something like that. Besides, I'm not quite sure people would like a robot feeding them, or the doctors wanting to risk the robot causing harm the the patients. Unfortunately, when it comes to medical devices, they can become VERY complex, expensive, and legal issues abound. Last thing we need is somebody suing CE for a robot that short circuits and starts shoving food down a person's throat. Lastly, consider that not many people need to be feed by somebody else, we wouldn't be helping a lot of people.

So basically, I think that a feeder robot is way too complicated for a project we all plan on doing in our free time that might not end up getting a lot of use. Still though, I really like the concept, but for now we should start on something we can actually build.

This is why I was thinking of a Fetch bot of some kind. Plenty of old, young, or sick people need help reaching down, up or around in tight corners or even moving around to get things constantly. Thus, we could produce something that is much simpler, and has a wider range of use. We also probably won't get sued for medical problems.

As for the image processing process, it could end up being a two part system. The first part could be like Ash said with a simple sticker / tag with a 2d symbol on it, the second part could be slightly more complex, matching the symbol with the object by itself. A Laser pointer could help the robot pick up non-sticker items or items that are to far away for the robot to ID.

Control could be a Automatic - Remote system. were a person gives a command to the robot and it goes off to fetch the device. Full manual control could be used as well to supplement the robot and the AI could keep track of all the times it was 'adjusted' and under what conditions so that it can learn from those inputs and then proceed to automatically preform them so the controller doesn't have too babysit the device.

Hm... just looking at this robot so far, it seems that in addition to control issues that programing is going to be key for this to be cheep. Main reason is that code can be copy-pasted without charge and that this robot is going to need AI to be of any real help no matter what it does.

Should we consider using a play station 2 controller (preferably wireless) to control the joint movements of the robotic arm ? Assuming the robotic arm consists of 3 parts:

1) Base
2) elbow
3) wrist
4) and a gripper capable of fetching objects, a claw perhaps.

The user can use the Play station 2 controller to control the DC servomotors located at the base, elbow, wrist and the gripper, resulting in the robotic arm movement to achieve tasks such as picking up objects. However the weakness is that the user needs to control the base,elbow,wrist and gripper one at a time.

I think we might be on to something, a controller similar in shape PS2 remote might be an idea. The buttons and control sticks offer plenty of control inputs. We could also take a cue from Nintendo and build a controller like a TV Remote.

well i think the controller shud be somekinf od gloves which will allow the user to give free movement of other arm,we can use flexible switches and a wireless connections for the control unit which shud be portable and easy to carry

First of This is a Superb idea Ash.

I say IN to this Project.
We need mechanical help also to design the blueprint of ARM design.

and as you said that Image processing we can use but i think as primary level lets try with Color Sticker or Barcode.

Immage processing is still in its infant stages,we cant get enough precision and accuracy!!!!,but we cant rule out that option

I think that by useing stickers and bar codes we could 'teach' the robot to better apply the Image processing system. Maybe we could use some kind of primitive radar as well to have it determine the composition of the object it is trying to pick up

When This arm is going to pickup the item?

i think for that we have to check the distance between the item and ARM when this distance is smaller than some predifined distance then after controller commands arm to bend the fingers and pick up the item.

hey guys,we are shifting towards Artifcial Intelligence,we will be attaching it to a human so we should be more concentrated on HMI,something which will be of no maintenance and extra-long life and durable tooo

sauravgoswamihey guys,we are shifting towards Artifcial Intelligence,we will be attaching it to a human so we should be more concentrated on HMI,something which will be of no maintenance and extra-long life and durable tooo

Sounds good Saurav 😎

How bout we build a robot with wheels which also has a mechanism for going up and down, either a pulley system or those big screw things similar to those robots in Robocon Govinda. Just this time we mount the robotic arm on a plate attached to the mechanism for going up and down. This results in a robot that can fetch things. The user can use a PS2 joystick to control the movements, without any AI.

Lock-OsMaybe we could use some kind of primitive radar as well to have it determine the composition of the object it is trying to pick up

I'm guessing anything capable of determining the composition of an object would be far from primitive (or cheap), Lock-O's 😛

With regards to Image processing, one doesn't need to do that many sophisticated techniques. Other than reading the barcodes and stickers, edge detection should suffice to get a sharp outline of an object, and auto correlate it with whatever is existing in the database.

JazlanHow bout we build a robot with wheels which also has a mechanism for going up and down, either a pulley system or those big screw things similar to those robots in Robocon Govinda. Just this time we mount the robotic arm on a plate attached to the mechanism for going up and down. This results in a robot that can fetch things. The user can use a PS2 joystick to control the movements, without any AI.

Yep, this would be pretty straight forward. For a manual control, the PS2 joystick will be useful. Perhaps we should start off with manual control yea? When we are confident with the movement of the joints and limbs, we can then move on to A.I. and all the image processing/radar jazz to cater with users with more serious disability conditions. Lets not leap too far ahead 😉

Jazlan has highlighted the main components of the arms in his recent post. Before we move on to the kinematics stuff, lets start creating some assumptions.

1. What is the maximum load of the object to be carried?
2. How far and high should the arm be extended?
3. Should the response be quick?

[spam]
Guys, I hope we don't get carried away with features. This robotic has to be simple & stupid.

Ash, did you hear from xHeavenlyx?

Yes, I propose we'll make the S & S version first then if successful, move on with the advanced features. I guess its okay to discuss the latter as long as we keep track of our progress.

After the e-mail, havent heard from him. He did say he wont have access to the net for a while. Poor dude 😛

sauravgoswamiImmage processing is still in its infant stages,we cant get enough precision and accuracy!!!!,but we cant rule out that option

:shock::shock::shock:

Well guys it looks like we were beaten to the punch. Some engineers here in the states already made a robot that pretty much does everything we planed it to do. Heck, It comes pretty close to my design I suggested, even using a laser to point to the target.

#-Link-Snipped-#

Maybe we could refocus on some other kind of task?

Cool stuff 😀

Hm, how about a low cost version of that one? 😉 That one is powered by a Mac Mini. I'm pretty sure we could at most use an embedded system.

Lets continue to target some simple goals before we move on. This project will still be a valuable lesson for all of us 😀

Come to think about it, this actully is a good proof of concept for us. replacing the Mac with a more basic circuit is a good idea. While the AI might be simpler, I think it's a sacrifice that can be made. So long as it gets the object to with the person's reach.

Hm, maybe after this we could move onto deciding the final layout and overall plan of the robot. I'd like to see our cheeper one roll out before the more expensive model in 2010. Heck, now THAT would be a challenge. Us doing open source V.S his team. If we can get enough dedicated supporters, maybe we could propose a challenge to them to see who can roll out their robot first.

Ah, 2010 seems like enough time.. *grin*

Game on!

Lock-OsWell guys it looks like we were beaten to the punch. Some engineers here in the states already made a robot ....

1. Man, there are Thousands of designs like this, all over the world, each better and more complex than the other. Its time to realise this: everything that needs to be done is already done somewhere by somebody else.

2. A project must be kept as simple as possible - the KISS method. This way, it has the benefit of economy and hence the probability of reaching more places of use.

3. Just MAKE something, no matter how simple it is. This will teach you several things.
Planning, in the name of design, always gets carried away out of bounds from the original desire and idea. Excessive planning will defeat the original idea and will end up as a non-starter.

4. A lot of project ideas I see here have already been done and have been around for years.

For eg: the current topic of this thread - Robotic Arm for assisting humans is the HOT favorite of all times, since the past 30 years. It has seen designs and implementation of thousands of them all over the world from school projects to Industrial products.

A prosthetic limb for the limbless is so advanced that it is controlled by the wearers mind (thoughts) through a sensor implant in the 'good' nerve plexus of the opposite arm. It is on sale for $25000 from the past 11 years.

5. Most of these designs are the from people who are NOT engineers at all.
The present day innovations are not restricted to Research houses anymore.

So, be realistic and keep things simple, else ideas have a weird habit of becoming a runaway fantasy rather than a realistic gadget.

This project will be fine, if you can restart from the first post.

Examine these facts:

1. Objectives of Cost and utility and purpose.

• A Human-assist arm is a complex piece of Electrical, Mechanical, Electronics & software. This ensures a high development cost. Servo motors, for example cost several thousand rupees each. A multi-jointed several DoF arm will necessitate atleast 6 high-torque servos. The example 'feeding' action means the arm must be capable of determining the food, the mouth and the relative positions of both. This will be impossible without some serious Image processing and accurate Kinematics, else the arm will dump the food in the persons eye.

It will also depend upon the mechanical rigidity of the arm. Any play in the system will make it impossible for it to position itself. A 1mm play at the base of a 1ft length will transform into a pendulum oscillating at a width of 2in or more. When this arm carries several DC motors and so much hardware, the wild swinging will simply dump the person to the ground HIM hurt him quite seriously. This calls for accurately machined joints. The cost of machining anything is -unbelievable.
​Utility can be as complex as plate to hand to mouth feeding or as simple as just moving the dinner plate. Mouth-feeding of food depends upon the type of food, in particular. This means, the 'hand' must be extraordinarily selective and dextrous to pick up and deliver, say, a piece of chapati or bread. This means that the 'hand' part of the arm must cut the pieces somehow, sometime, somewhere.
​

• Or to open the door.....now - why do I need a Robot to do that? it is simple to install a Solenoid lock that can be operated remotely with a shout.

The Arm MUST to be both Autonomous & Remote controlled. Autonomous means sensors and their intelligent management. It calls for a lot of Embedded software.
​It has to move around the house and so Navigation and path planning is a must. This means motors powerful enough to move the Robot. It also means a big Robot heavy enough to be absolutely stable on the ground.
​All this calls for a Heavy Robot and hefty batteries to keep it running.

Looking at my book of estimates, the first working prototype will cost upwards of Rs. 3 lakhs.

Now a simple suggestion:
Its time for people to stop planning too much and start making the Arm, no matter how simple, crude, ugly, stupid or useless it may be!

Theres' been enough talk but NO work at all till now in the Forums.

Good luck !!

Well, I'd like to work on it, but talking and coming up with ideas is good and fun as well. Still, unless it leads somewhere, it doesn't serve us any good.

While I don't think it right for me to suddenly stop generating ideas that might make it's way into more advanced models or later projects, I'd like to start the prodject in the near future.

Hm... while we could just post all of our plans here, we might get distracted and wonder off topic. How about we hold a design contest like they do to determine the best device? We can all submit our ideas to a few judges and have them decide what one will become the CE prodject, then from there we can finish up all the details as a group. That way we skip the earliest parts of the design so we can gain some focus and bypass weeks, if months, of idle, wondering debate.

The prize winner will be head of the prodject, and at the end of it all he'll be able to put the prodject down on his resume as the reward, something that is more valuable than cash these days. I'm thinking that maybe 2-3 weeks will be enough time to submit a rough overall plan to the judges.

If you guys want to do it that way, then we need to come up with a target goal, so far, we seem to really like the idea of a fetch-bot of sorts, so let's set that up as our objective. A robot to pick up items on the ground and move them to a sitting person's reach.

How does that sound to everyone?

The contest idea is interesting, but to say that the winner must lead the project may prove to be a bit of a burden for him/her 😉 Well, regardless of contest or not, is there ANY one who would like to volunteer to lead this? He/she doesnt need to be the design expert.. just someone who can facilitate discussions and ideas. The crux of collaboration is that all ideas is shared and deliberated upon.

Docel, its apparent you have vast experience in this type of projects. Perhaps you can give some design ideas?

We don't have to start from the beginning. There are some nice ideas in this thread already. Jazlan, for example, has contributed some reachable ideas in which we can follow.

The image processing stuff can be saved for later. AT least, if we ever reach that point, we can quickly dig the info back up 😉

So, our objective remains the same. We want to have a robot to assist the human somehow. Lock O's has said, perhaps a fetch bot:

A robot to pick up items on the ground and move them to a sitting person's reach.

Maybe we can still extend it to a table surface as well. But, lets deal with assumption questions I mentioned in one of my previous posts:

1. What is the maximum load of the object to be carried?
2. How far and high should the arm be extended?
3. Should the response be quick?

After those assumptions, and finalization of objectives, that would narrow down the design scope. What I'm sure people would love is a darn sketch or diagram 😛

I'm game, Ash !
I always start things from scratch and move up to improvisations. That way people start making something. This will lead to a lot of practical skills and the experience eggs you on to improvise. Most students havent done anything practical independently till now, simply because there was neither a chance or opportunity or motivation to do so. In time, they hesitate to start anything by themselves.

So: just give them an opening......

A construction project does the job, provided there are consistent Forum visitors. The visitors are sporadic due to various reasons. One of the reason is the academic Project time. That is when they are suddenly active and flit around trying to find a project they can do. A couple of weeks and they fall prey to unscrupulous 'ready made project' vendors and disappear.

Now, the Project time is approaching and the guys are dabbling around with ideas between the AUV and the Mars rover.......

Before they disappear, lets give them a hands-on construction project so that they hang around!!!

I can lead the way with a Build project, if I have a Quorum here.

Instead of using expensive servos, you should consider using shape memory alloys (like Ni-Ti alloy). A typical wire costs between $ 2.5 - $ 9/ft (Source - #-Link-Snipped-#). So even if you have to use 10 feet of wire it works out to be $80 roughly.

Some robotic arms based on SMAs

#-Link-Snipped-#
#-Link-Snipped-#
#-Link-Snipped-#
ieeexplore.ieee.org/iel5/6536/17442/00804917.pdf?arnumber=804917

mad_scientistInstead of using expensive servos, you should consider using shape memory alloys (like Ni-Ti alloy). A typical wire costs between $ 2.5 - $ 9/ft (Source - #-Link-Snipped-#). So even if you have to use 10 feet of wire it works out to be $80 roughly.

Some robotic arms based on SMAs

#-Link-Snipped-#
#-Link-Snipped-#
#-Link-Snipped-#
ieeexplore.ieee.org/iel5/6536/17442/00804917.pdf?arnumber=804917

😀

Yep, I did think about that for a while, but SMA's are an overkill for our first prototype. SMA's in robotic arm application are usually used to closely mimic human muscle movements.

Servos are good enough for us. Personally, I feel SMA requires more research and time before we can implement them successfully, which includes figuring out how to accurately bend them under various environmental conditions (temperature, etc).

But, thanks for the info. We'll keep SMA for future use. It's easier to design the arm around servos and later upgrade to SMA than the other way round.

*bump*

Any volunteers to design a simple robotic arm? Google SketchUp is quite easy to use 😀

I saw something similar a few weeks ago on television in which the human arm was replaced by a robotic arm and all the joints and all were artificial and they moved with the smae speed and control that a normal arm would have.

This thread is awesome. way to go friends, gonna go through all the pages in detail in a minute. 😀

Haha, thanks for bumping this thread.

We want to start off with something simple first before we jump into the realm of human prosthetic robot arms.

What we need now is some sort of basic design to look at. I'm a bit tired of on going discussions without diagrams and pictures 😛

Again, we begin with a robotic arm that can be remotely controlled, and can pick up an object. Arm will consist of the base, the elbow, wrist and finally the gripper.

Later, we can add autonomous features and sophisticated image processing means. But I stress "later" 😀

ok guys only discussions here or any work has been done??? I'm making one for a friend's project.. any worthwhile guidance ?