Well, sort of!!! I do appreciate your time-consuming efforts to educate pork! ? ?
But I still don't see where the OpAmps fit in. Wouldn't it be simpler to start with smilies, with different facial expressions, on bricks placed around the robot, for initial testing purposes, instead of actual complex human faces?
Classify the expressions and then tell the robot to work its way around the room until it found "happy" for instance.
This I believe would certainly be an achievement for the type of robot that Bill is building. Your proposed venture sounds very much to me, like a project for a science and engineering institute, with a large budget and some real computing power, rather than a small cellar hobbyist workshop.
But I will not dampen your spirits and look forward to any video you may produce on this subject.
This is what I mean about it being both simple and complex. I'm actually working on two different aspects of this at the same time. One aspect is a totally new and unorthodox approach to training neural networks. And that can still be done via a computer simulation as is currently being used by just about everyone who works in this field.
The second aspect is to then take the finished modeled ANN and convert it into a hard-wired analog circuit using OpAmps. What's the idea with this latter step? Well this transforms the neural network from a computer generated simulation to an actual physical analog entity that is much faster than any computer could ever hope to calculate. In fact, this physical ANN is actually far more representative of how our brains actually work. Our brains do not simulate ANNs like a digital computer does. Instead, our brains are actually hard-wired physical networks that are basically constructed of biological OpAmps.
So I'm basically moving the ANN from a computer simulation toward an actual physical "brain".
Having said this, I could just simulate my ANNs in a digital computer like everyone else. But this then requires that I continually purchase faster and more powerful computers. Digital computers are actually extremely inefficient. What makes them so useful for us is the ease that they can be dynamically programmed. But in truth, analog computers are far more efficient when it comes to actually computing power. And this is what a physical ANN would be. Like our brains it would be an analogy computer instead of a digital computer. The later can only simulate. The former is the "Real Thing".
Wouldn't it be simpler to start with smilies, with different facial expressions, on bricks placed around the robot, for initial testing purposes, instead of actual complex human faces?
Absolutely. I only gave the far more complex example of recognizing individual human faces as a hypothetical example. Plus there is far more to it than I can possibly explain in these posts. This is why the need for entire video lectures. If I could explain it clearly in a post I'd just do that. But that's simply not possible.
This I believe would certainly be an achievement for the type of robot that Bill is building. Your proposed venture sounds very much to me, like a project for a science and engineering institute, with a large budget and some real computing power, rather than a small cellar hobbyist workshop
There is no question about it. To accomplish the ultimate goals that I would like to accomplish I would actually need a science grant and a team of dedicated ambitious college grads working with me in a team that is all focused on the same goal.
That would be extremely nice to be sure. No doubt about it.
Exactly how far I'll get as a lone hobbyist is impossible to say. This is another reason I would like to make videos to explain what I'm doing. There are a lot of people working on ANNs and whether they choose to go down the same path I'm looking at with hard-wired analog ANNs, or even if they choose to continue down the mainstream path of digital computer simulations, they would still find much benefit from the info I'll be including in my videos.
In fact, I'm planning on making at least 3 introductory videos just on explaining how perceptrons work. There won't be any mention in those videos about building hard-wired OpAmp ANNs. That will only come later after the behavior and modeling of ANNs is first well understood.
So this will be a very slow moving introduction for those who have lots of patience. ?
I will also be starting at the very beginning with a 2-input perceptron. My videos will be accessible to everyone even if they never heard of an ANN and have no clue what it does. By the time they have watched my first 3 videos (assuming I ever actually record those videos), they will have gone from knowing nothing about how ANNs work to fully understanding how a 2, 3, and 4-input perceptron works. Each one becoming 1 spatial dimension more complex than the last. The 4-input perceptron takes us into 4D hyperspace. And from there we'll be far more prepared to address arbitrary nD hyperspaces.
And these will, of course be hyperspaces of information, or data analysis.
The example I gave of recognizing a human face is actually a simple example. One that we can easily visualize because we, as humans, know what it means to visually recognize a face. But that's just one example. The other area I'm interested in is the hyperspace of abstract concepts to be used in Semantics A.I. That's a little more difficult to understand unless a person also has a good foundation in Semantic A.I. And I would like to eventually make some videos on that as well. But let's take things one at a time for now.
I don't know if I want ai in duce but some would be fun!
AI is not like adding a single function to your robot it is a label for particular types of processes some which are very simple to those we have no idea how to implement such as a human brain. You need AI of some kind unless your robot is nothing but a remote controlled toy.
When we label a behaviour as intelligent we are observing some kind of goal seeking behaviour. It can be as simple as seeking light to seeking a win in chess to something more complex.
AI is not like adding a single function to your robot it is a label for particular types of processes some which are very simple to those we have no idea how to implement such as a human brain.
When we label a behaviour as intelligent we are observing some kind of goal seeking behaviour. It can be as simple as seeking light to seeking a win in chess to something more complex.
Exactly. There are many different levels of AI. Basically most modern day robots have some level of AI capability. If your robot can find its way around the house on its own that's already a level of AI.
Posted by: @casey
You need AI of some kind unless your robot is nothing but a remote controlled toy.
I wouldn't go quite that far to say that a totally remote control robot is nothing but a toy. There are many practical uses for remote control robotics. The most obvious would be a robot that is under the control of someone who simply cannot physically do the task the robot is performing. So even though the robot is under full remote control of the operator, it's still much more than just a toy.
I'm currently dealing with a huge oak tree that has fallen toward my house. I need to climb up about 40 or 50 feet in the air and cut off some limbs. I also need to get up on the roof of my house and cut off limbs that have already reached the roof.
Right now, I'd give anything for a totally remote control robot that could take a chainsaw up and cut those limbs off for me. The fact that I would need to control its every move would be fine with me. That still much better than me having to climb up there with a chainsaw and cut those branches off myself.
Unfortunately I don't have a remote control chainsaw wielding robot. So I'm going to have to take my biological robotic body up a ladder and do it myself. Live in-person with gravity tugging at me to come back down in a big hurry! Something I really don't want to do.
But yeah, even totally remote control robots can be quite useful.
Yes remote controlled robots can be more than a toy I was just thinking of some of the robots I have seen on the internet that are essentially remote controlled toys. I should have said a remote controlled robot vs. an autonomous robot.
I have a similar problem with pruning a very large tree and day dreamed about a tree climbing robot carrying a chain saw. Do I pay to get professionals in with their special elevator platforms and expertise or try and do it myself? I recommend getting in the professionals. Cheaper than falling down and ending up a bag of broken bones. Just one slip up and, like someone I knew, the trunk comes down or bounces up giving you a blow to the head and that is the end. He was only 30 with a wife and two young children.
Looking at the image you posted it appears the tree has landed on another tree. If it isn't going to fall onto the house itself maybe just leave it as is. A professional should have a good idea of the safest way to bring it down.
Duce could never be a remote control robot its to big and it has the infrastructure in place for ai the only remote controlled part of duce is the cover for the 3d printer.it should at least navigate a room it does have ros on windows I know no one is a fan of that the one caveat but hey it has the panda to run ros it has pi 3 for fun pi 4 for the 3d printer it has its relay hooked up to the ir sensors front and back and all of its ultrasonic Senors front and back working just needs about 200 thousand hours of programming and figuring out motor code is going to be fun also I need to get on that barely ran over the edge of my toe with it last night and believe me at 235 lbs a toy it aint
Do I pay to get professionals in with their special elevator platforms and expertise or try and do it myself? I recommend getting in the professionals
I would love to hire a tree surgeon to take the tree down. Unfortunately for me that's just not a financially feasible option. We'd be talking close to a grand, and I just don't have that kind of dough to pay someone else to do it. Even half a grand would be out of the question.
Posted by: @casey
Looking at the image you posted it appears the tree has landed on another tree. If it isn't going to fall onto the house itself maybe just leave it as is. A professional should have a good idea of the safest way to bring it down.
One could say that I am a professional when it comes to taking down large trees. I don't get paid for it, but I've been taking down 60 foot tall trees for years on a regular basis. I do it both for winter fire wood and for use to make into lumber on my sawmill:
So while I don't have the fancy elevators, I do have tools and methods for dealing with huge trees like this. Thus far I have already brought this tree downward a good 10 feet. I don't have any current pictures but originally it was still on a 10-foot high stump. I've cut it off that stump and the whole thing has fallen 10 feet lower. Lots of safety lines to prevent the tree from accidentally rolling or falling onto the house. In fact, if the house wasn't there it would be a piece of cake to take it down. The real problem is that it wants to fall on the house and so I need to use a lot of strategically placed ropes and chains to prevent it from falling in that direction. So I'm 'inching" it down a little bit at a time. I'm actually pulling it down using chains and winches attached to the trunk. But again, because of the house being under it I need to do this one inch at a time.
When I cut branches off up high, there is no chance that the tree will actually fall during that process. It's very well stuck in the "Y" of another very large oak tree.
Anyway, I just thought I'd add that this isn't like something I'm trying for the first time. I've been doing this sort of work for many years. So I have a good idea of what the dangers are and how to handle them. Still, I don't like climbing ladders to even do easy roof work anymore. And there is, of course, always a chance that something can go wrong. Just because I know what I'm doing is no guarantee that it's safe. ?
But I'm just saying, I'm doing it as safety as possible. But heck yeah, if I could afford to hire professionals to do it for me I'd do that for sure. That's just not a financially wise decision for me at this time. Taking it down myself includes risk, but it's also costs zero dollars and zero cents. So that's my reality. Besides I'd rather spend a grand on robot parts if I had that kind of money. ?
I don't know if I want ai in duce but some would be fun! Maybe it will figure out how to print before me lol probably . ?
When I saw the first picture you had posted of Duce I thought the head was like a giant frosted-glass light bulb shape. In this photo it appears to be just a flat sheet of acrylic or something with a face painted on it?
It's an interesting robot to be sure. Quit large. The arm and grasping hand appear to be quite large as well. It would be interesting to learn more details of the actual build. How did the idea start? What are its current capabilities. Looks like you could make quite a few videos explaining how Duce works and how you've built it.
The head is 5 in thick from face to back with mylar wrapping it the face is an insert that I cut out and put in between 2 pieces of lexan with the LEDs behind it inside that's what makes it appear to be painted on in pics but there is an uno inside mainly for the remote control for the 3d printer cover some of my build is on YouTube under duce robot build I'll be posting more detailed video soon but if you have any questions I'll be happy to answer them the duce has been a 4 year project but I have been milking it because its so much fun quote its not about getting to the top of the mountain its about the climb ??
its not about getting to the top of the mountain its about the climb ??
Exactly. It's a hobby. As long as you're having fun it's a successful hobby. ?
I didn't know you already had videos up. I just watched one of them. I can see where you are really enjoying all the different aspects of your robot. I like the idea of having built-in dioramas for special effects and artistic interest.
I have big dreams, but you have a big robot. So you're way ahead of me.
Thanks ! It's certainly a labor of love I wish I would have gotten in on this stuff in 1984 when I really had the chance of course I was 16 then but I'll be asking a lot of questions to get this robot finished so get ready I'm getting ready to start ros and the 3d printer I don't know if you can see it but there is a 3d printer mounted on it so that will be an adventure its almost ready though I'll get some video of it doing something soon . ?
I wonder if these could be used to let other people join in on this project to contribute rgb sketches and such even maybe ai there are so many rgb boards in it .it might be fun for people to see they're handy work put right to use let people be the base for it the more people the deeper the ai the Borg theory like you said maybe you could come up with an idea for it ?
it might be fun for people to see they're handy work put right to use let people be the base for it the more people the deeper the ai the Borg theory like you said maybe you could come up with an idea for it ?
The best way to get contributors is to post a schematic of what you already have. They could then offer suggestions, code, sensors, and wiring from there.
You could start with a simple block diagram of all the boards, sensors and motors you already have installed.
Something generic like this:
Only be a lot more specific in terms of exactly what each board is and try to include the exact numbers for motors, sensors, etc.
The connections between everything doesn't need to be detailed yet. Just get the general idea of what each board is controlling or connected to.
Then as questions come up you can provide more precise details of exact wiring connections pin-to-pin etc.
The more information people have about your system the better they will be able to make suggestions.
We use cookies on the DroneBot Workshop Forums to give you the most relevant experience by remembering your preferences and repeat visits. By clicking “Accept”, you consent to the use of ALL the cookies.
This website uses cookies to improve your experience while you navigate through the website. Out of these, the cookies that are categorized as necessary are stored on your browser as they are essential for the working of basic functionalities of the website. We also use third-party cookies that help us analyze and understand how you use this website. These cookies will be stored in your browser only with your consent. You also have the option to opt-out of these cookies. But opting out of some of these cookies may affect your browsing experience.
Necessary cookies are absolutely essential for the website to function properly. This category only includes cookies that ensures basic functionalities and security features of the website. These cookies do not store any personal information.
Any cookies that may not be particularly necessary for the website to function and is used specifically to collect user personal data via analytics, ads, other embedded contents are termed as non-necessary cookies. It is mandatory to procure user consent prior to running these cookies on your website.
