The best way to have real time video processing and use the data to make your bot do neat things?
Perhaps it’s best to use a wireless video device to send the data to a remote computer, process the data, send the results wirelessly back to the bot for control. A wireless receiver would be needed to interface the data feed from the computer to the bot’s microcontroller running custom code to send servo commands.
There is no real tutorial on how to do this for those wanting intelligent autonomous control using this method. Right now, you can use RoboRealm to process the video and send commands directly the SSC32, but there really isn’t any intelligent decision making. Basically servos will move left or right depending on the image it sees. My goal is to have the microcontroller choose what to do based on this data.
mike, nick ,and everyone else thanks i knew i could count on you for help
believe it or not that is a great idea and a lot of the componets are pretty
cheap but my computer that i do robotics on cant run robo realm this probally because i dont have the net on it as well as e machaine shop
but the question is would it be possible to use 4 hs805bb servos for my
project also where could i get micro processors
oh yeah where can i get brackets for a hs805bb servo
Some microprocessors can be found at Lynx here: lynxmotion.com/Category.aspx?CategoryID=84 This doesn’t include PICs or some of the more exotic micros like 8051 or Propeller, nor embedded computer systems like PC104 boards. You might to check out Parallax’s web site, and general internet search too. As far as brackets, the SES doesn’t have any Mega (1/4 scale) servo brackets. I think you’d have to make your own there. Nick, maybe you can enlighten us as to what you do to make the Mega bracket in your biped?
Personally, I’d love for Lynx to sell a Mega scale version SES so I won’t have to worry about fitting my electronics on a bot, I could just build a bigger one. I dub thee MegaScout! I’d also love for there to exist a Mega servo which scales to the power of a 5955 (around 650 oz/in of torque), but that would be a ton of work on Jim’s part to design and get the large scale brackets fabricated. It also would probably be prohibitively expensive for most people to work at that scale as some industrial servos are roughly that size/ power but cost around $250 a piece. The current needed to supply those servos would also be truly massive, further complicating matters.
good news i’ve just got the book today this actually answer some of my questons i also got the books pic robotics and how to build a humanoid robot but i’ll still need help on the robot vision idea
I can help with the robot vision, no problem. I still think you would be better off starting from square one. Must of us here would like to dive into everything at once and some of us have done that before, but have paid a price for it.
Glad you got the book, it kept me busy for days. I’m still using the BS2 Homework Board and trying to program new things as I get ideas. I used this same board to develop my sensor board idea that evolved into the RSCU board. I had to take the needed steps to get where I am now, and there is a great deal of new territory I have yet to venture into.
Nick has taken the same rout where he started small and progressed to a guru know-it-all, show-off… that he is today (JK Nick).
Anyway, I would start out small and build on that knowledge.
I just started reading this thread and it looks very cool so far!
I noticed a few people recommending the RoboRealm software. I would agree with them and say to go this route. RoboRealm has some pretty nifty features that make vision processing really easy. I created a setup in less than a few hours that made a pan/tilt camera track the flame of a candle. You can even train it to recognize certain objects! It only requires knowing VB script, which is very easy to pick up on. You also arent required to use a microcontroller if you would like to save some time.
Agree! RoboRealm is becoming more powerful with each new release. With the latest version there are now more features than I could possibly pack in a single pipeline.
VB Script is one thing I keep forgetting to mention when I talk about RoboRealm. I’m not much of a programmer but for someone who can program using VB Script can utilize the full power of RoboRealm.
I wanted to have color vision for my biped, but I’m thinking that it’s going to be to much for it. Not real sure, we’ll see.
I would like to also make a “Johnny 5” type bot for advanced robotics in the future because it has a nice “truck bed” for all kinds of electronic goodies, with out much worry for weight since it is on wheels.
thanks mike the book is working wonders now i got how to build my biped
first the basics thats understanding the ssc-32 and servos and brackets this stamp is amazing
The stamp is a nice controller. Its like a small computer that can run basic code. The stamp is not as fast as the Atom micro, but for simple robotic stuff it’s very good. A biped would need more speed however, such as the atom, to be able to run larger programs, which is what I plan to get next.
You plan to make custom brackets for your biped? This can be very challenging.
yeah but i have a plan first i’ll ues sheet metal to prototype
then aluminum you see i can get sheet aluminum almost free
…oh yeah which is better hobbico cs72bb or hitec 805bb servos
yeah i’ve just checked and you are right one more
queston whats the differrence between digital and analog servos
oh i found a hitec 5745mg servo but boy it’s high cost
sorry one more queston what can i do to get a discount on my servos
because i’m buying all my servos in one lump sum
If they’re from Lynx, you can goto the servo deals page: lynxmotion.com/Category.aspx?CategoryID=93 but you have to buy in multiples of six. If you’re buying a lot but not a clean multiple of six I think you might be able to work a deal with Jim. I know he was real nice when I ordered my 5955s (before the combo for the 5955s was online) giving me an automatic discount of I believe 10%. I was a bit surprised at first when I first looked at my packing slip, but in a happy kind of way!
Why would color be too much? Since you are building a biped and plan on using RoboRealm so I assume you will have the video transmitted back to a PC, is this so? Are you going to have an onboard MCU to control the basic movement and then have the PC process higher level tasks? What sort of things do you plan to do with the vison processing? Sorry if this has already been mentioned but I have not read the thread entirely yet.
Just wanted to correct myself while I am still thinking about it.
The aluminum or the bridge of the nose of the bot head was not the main cause of poor performance in my bot head design. The reason it did not work fully sunk in the eye sockets, was due to only a portion of the sensors having a barrier. The sensors will work fully inserted in the eye sockets as long as a short piece of thin wall aluminum tubing is used for both sensors like Jim used in that picture a few posts back, only he used PVC instead of aluminum. The key is a symmetrical barrier around the full circumference of the sensors as well as equal lengths, but not to long. I found this out by taking a piece of construction paper and making my own “tube” to put over the eyes and it was reading like it should.
I have wanted to post this but kept forgetting to. I think I am going to have another bot head made with a few new improved modifications. The current bot head I have has a few boo boos. One was when I took my dremel and tried to buff it out a bit and the chuck took a bite out of the jaw area, also I ground a bevel on the bridge of the nose to try to improve the sensor performance. It helped, but is not satisfactory. I also was successful enough to put a nice gouge in the sensor. All these are small imperfections, but seeing them is like walking around with a stone in my shoe.
Now that I have a better understanding, I want to do another one.
I put this on the back burner for now. I was a bit unhappy with the way the PING sonar was performing and got frustrated trying to get it to work. What I have discovered is that for it to work well, the sensors need to be shrouded EQUALLY around the full circumference. In other words, you can’t have one side taller than the other (bevel cut). I would need to get a piece of aluminum tubing, and cut both pieces squarely and of equal lengths.
My problem is the bot head has the eye sockets beveled. The Bridge of the nose is causing interference. A piece of aluminum tubing the with the right O.D. can solve all the problems but then I would want to remount the PING sensors deeper in the eye sockets as I had originally wanted. Problem is, that industrial epoxy wont let go.
I would like to have another one made and do it right. For now I’m taking a break.
if you are willing to sacrifice the ping board you can probably heat the skull up enough in an oven to get the epoxy to let go. I wouldn’t use a blow torch though because it will distort the skull as it heats very unevenly. make sure you have the hood fan over the oven on 'cause it’ll be stinky.
I’d also love for there to exist a Mega servo which scales to the power of a 5955 (around 650 oz/in of torque), but that would be a ton of work on Jim’s part to design and get the large scale brackets fabricated. It also would probably be prohibitively expensive for most people to work at that scale as some industrial servos are roughly that size/ power but cost around $250 a piece. The current needed to supply those servos would also be truly massive, further complicating matters.
(JK Nick).