Let's Make Robots!

New and need some detailed advice

Hello, i'm new to LMR and Robotics in general. Some of you i'm sure have played TF2 and know about the Sentry Bot. and Also, you may know that Weta Workshop have recently built a working (kinda) replica for VALVe.

 

Now, i can build this, and most likely it will function correctly. However, i need to know what sensors, speakers, and god knows whatever else i will need to get it to Track and Follow my 'target'. Also, will need information on what servos/motors would be needed.

 

Thanks in Advance, Ghillie

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

This is just a "big picture" vs "little picture" issue. The idea that there is "a way to code this" is like saying "I need to know how to operate a machine". Let's break this down a little bit.

First off is the movement itself. Let's assume you can find servos big enough. You will need to decide how and when the position of the servos is going to be updated. Will you step from position to position or simply allow the servos to go full-speed to their position? If they are going to get from A to B in an incrimental way, is it going to be proportional? Bigger steps when you gotta travel faster? If you let 22lbs go full speed, can you get it to stop? If not, you will have to figure in a ramp-down in speed before the servo gets to its destination. All this will need to be experimented with to figure out what works. --A lot of time will be spent just getting the mechanics working.

Now, we have tracking. The most basic will track edges, not objects. A distance sensor would be used. This distance sensor will be swept back and forth and up and down. As it scans, each reading is stored in a variable so it can be compared with the next reading. If the difference between reading A and reading B is greater than a given threashold, we can assume it is the edge of an object. We repeat the procedure by using a second sensor, or sweeping the first sensor to where we "think" the other side of the object would be. Again, we sweep and try to find the other edge of the object.

After we have determined the total width of the object, and compare it to the mean of the R and L distances, we can get an idea of the actual width of the object. At this point, using a look-up table, we can compare our width with the know width of say, a human. We now now we are looking at something, that has an "edge" on either side, and is probably about the size of a person.

From there, we can find the midpoint of the R and L edges and then command our servos to point that direction. This whole procedure must be doubled if we tilt as well as pan.

Now mind you, this is just a 30,000' view of what needs to be done. Everything decribed above would need to be taken apart, and coded piece by piece before it can all be put together into the final product. There are 1000's of little things that must be learned to be able to put these "little things" together into something cool.

Don't get frusterated --this is a great hobby and this IS something you can build. Go buy an Arduino or a Picaxe and get a LED blinking. I promise --once you get the LED blinking, you will understand why this is so complicated.

 

Thanks, this is exactly what i was looking for. I think the sensing you described is a little overcomplicated, but hey, its still going to be a massive problem ;D

I'm getting in contact with somebody from a nearby Uni and i'll see if he has any advice. If not, i'll dump the sensing for now, and start small time :)

Thanks again everyone for your help :)

 

Oli

If you are new to robotics, this is a bit much to get started with. What you need now is a microcontroller, a breadboard and a LED. Start with getting the LED to blink and go from there.

I really am trying to save you here. In the post above, you mention a 22 pound head and servos. To be able to move somthing 22lbs at the speed needed for motion tracking is going to require massive servos/actuators. If you are actually going to try something like this, with the dimentions described above, you will be looking at 1000's of dollars and (with the learning curve of code) a year or two to get it working. I just want to keep you from having 2000 bucks of parts lying around while asking questions about for/next loops.

LED... Microcontroller... Blink...

$100 to get started.

I appreciate your concern. I realise that this is a huge undertaking, and i know its going to be tough. I really came here for an idea of what is required to do this, in fair detail. Such as how to code something like this, and the set-up i would need. I know the set-up, now i just need to know what the code would have to be like.

And it isnt required to move at massively high speeds, more like 120 degrees in a second, probably more.

 

I know i sound like an arrogant tool, but i don't mean to. I just want to find out as much as possible so that i can determine whether going all the way with this is a viable option in terms of funds.

 

Oli

I have no experience in this really, i just know what the things do, what i want them to do but not how to do it.And in reply to your questions i'll post up the sizes etc

 

I think it is going to be about 18 inches by 24 inches footbase and 25inches high. The base can be considered static and will contain the majority of the weight, but i'd estimate that with Aluminium and Weight reduction, the head will wiegh approx 10-12kg. As for the servo size, it has about a 1 1/2" tube in which to operate or that is the area in which mechanics can go to allow rotation. As for the electronics guff, there is a 6" and 4" deep drum at the back which can house most of the electronics.

I just want to thank you again for all your help, it wouldn't have been possible for this project to start w/o you guys :)

 

Oli

If you’re planning to build this at full size and in steel, then the weight will be much more than 6kg. If you look at the video 4 grown men are lifting it and I don’t think it’s because they are afraid to drop it. You’re probably looking at more like 50kg.

And if you look at the complex metal work on the base it would probably take you weeks just to machine it. So if you come back with a complete base and a rotating ‘head’ we could probably help you with the servo / stepper and sensor, but I think that is the least of your problems.

Thanks for the reply :)

Take into consideration the majority of metalwork is on the base/struts and feet. also, i'll cut down material to reduce weight. But i appreciate your help, and i will return when i have manufactured the Beast xD

Thanks, Ghillie

If the plan is to give it to your school, you might also consider if this is something a school would like to have in its hallway. The kids might think it’s cool to have a simulated weapon tracking them, but I’m not sure the teachers would think the same.

I do know Valve and enjoyed the many hours I spent playing TF2, but if you are a beginner and plan to make something similar to what Weta did, that is an incredible undertaking. GroG's idea, I think, is probably best for this, although you might want sensors with a Bit more range than 2 metres. Have them mounted on top or in the front, one straight ahead, and the other two either angled slightly to either side, or sitting a distance to either side, and point straight ahead. Have the turret, when it detects an object within range, as GroG said, aim the centre sensor at it, and then if object moves to either side, the final sensor it passes by (left or right one), have the turret turn in that direction until object is seen again. Repeat. This can easily be acheived with a small microcontroller, but what might be a problem is getting and powering servos to turn it. How big are you going to make this? And for sound, just a simple beep and shooting sound, a small, cheap piezo would probably work. Failing that, a small, cheap speaker.

Well, it is going to be a meaty machine, full steel, turret should weigh in at ^6kg

And would all this be possible w/o a Computer?

thanks, Ghillie

PS, if i decide to do this, would it be ok to send a PM to you, so you could give me some help with it?