Society of Robots - Robot Forum
General Misc => Robot Videos => Topic started by: sonictj on July 24, 2009, 05:03:35 PM
-
This platform is based off of my atmega640 prototype PCB I created a little while ago. This platform uses 6 s148 Futaba servos (free courtesy of my father an RC hobbyist). The three wheel servos are modified for 360 degree rotation. The platform has the ability to move in any direction while maintaining a consistent heading. I designed this bot to be a little bit showy, so that I can bring it to an interview to demonstrate my skills.
The video shows the robot doing a counterclockwise circle. You will note that the robot pauses to rotate individual wheel assemblies. This is necessary because the steering servos have a maximum rotation of 180 degrees. I also intentionally slowed the spinning to put less stress on the servos. Oh, and the bot is orange and blue because I go to UF. Let me know what you think, and also if you can think of a good name.
[youtube]DPz-ullMOqc[/youtube]
-
nice job sonictj, cannot think of a name as thinking of names has never been one of my strong points.
Just out of curiosity what material is the main orange and blue structure made from. ;D
-
the structure is .25" oak plywood. Its just what I had lying around. I wouldn't recommend it because of how porous it is. I had to put around 5 coats of paint on the thing. If your gonna use wood I recommend Masonite. Masonite does not splinter as easily and its also very smooth like plastic.
-
Reminds me of my Taurus 2 ;D
(but yours looks much better designed, as I was a noob way back then)
-
very nice, i like it. i don't suppose there would be anyway to get around the stopping to spin a servo unless you went to gear motors with encoders for spinning the wheels. You kind of have to rely on the position control offered by a servo right?
-
very nice, i like it. i don't suppose there would be anyway to get around the stopping to spin a servo unless you went to gear motors with encoders for spinning the wheels. You kind of have to rely on the position control offered by a servo right?
I may go to steppers eventually, but the spinning of wheels is unavoidable even with a 360 degree setup. The wheel motor power cables can only be twisted so much. I decided to go with servos because I wanted to decrease the amount of circuitry on the robot.
-
The rotation servos don't need to overcome much torque. Considered using high speed servos instead?
-
The rotation servos don't need to overcome much torque. Considered using high speed servos instead?
I've already ramped up the speed. I was intentionally being conservative at first to not break anything. I don't see much reason to spend the money to get the faster servos at the moment. I like free servos ;D. Thanks for the suggestion though.
Oh, and the bot should remind you of Taurus 2 because thats where I drew the concept from ;).
-
I like it though I fell they're should be a gripper or something on the end
-
I like it though I fell they're should be a gripper or something on the end
I'm adding stuff for sure no gripper planned yet. What you guys saw was the finished base and nothing more. At this point I've created a custom LCD mounding bracket to hold my 20x4 display, a 180 deg sonar turret, Sharp IR x4 for wall following/obstacle avoidance, and a second sonar for the rear of the robot. My plan is to use a wii remote again for tracking IR hot spots. I also have reflective sensors x6 that I may add on for line following behaviors. I'll post some cad drawing in a minute or two.
-
Here is what I've CADed up. My robot has all of these features now too. I just finished making/fiitting the parts today. I'll post more pics and video soon.
-
nice cading, the best and clearest i have seen in a long time. ;D
-
this robot really looks like its going to be a success!
well planned out, professionally made parts and great dedication...
great job sonic, i hope to see those cads become reality...
-
i hope to see those cads become reality...
Here is the pic.
-
Here is the latest video demonstration of the robot's mobility
[youtube]hxUIS3_RGGY[/youtube]
-
very nice, with the turning of the individual wheels sped up, it makes up for the fact that they have finite rotation. With the being fast, the robot having to pause for a second doesn't seem like that big of a deal.
-
very nice, with the turning of the individual wheels sped up, it makes up for the fact that they have finite rotation. With the being fast, the robot having to pause for a second doesn't seem like that big of a deal.
yeah I think watching the robot position the wheels is the best part. I set it up so they cascade too. This prevents the robot from shifting too much from all the wheels moving at once.
-
the new guy looks great!
i really like how you sped up the turning of the wheels, although watching the robot turn its wheels does look kinda cool ;D
-
the new added hardware looks great sonic. can't wait to see what happens next ;D
-
This is another video of my three wheeled omnidirectional robot. This video shows the robot ("
GatorKS") following the wall of my kitchen.
First you will see the robot align itself parallel to my kitchen cabinets. Then the Ping)) sonar is used to position the robot a set distance from the wall. The robot then runs along the wall trying to maintain that distance. The Ping)) sonar is used when the robot encounters sharp turns.
The way the thing drives is servos A and B pivot, while C stays parallel to line AB on the robot frame.
setup:
A
wall / /\
/ / \
/ / \
/ B______C
two Sharp IR sensors are used for the calibration with the wall, and then only the front is used for the wall following. The one IR sensor's initial value taken, after the sonar spaces the robot from the wall, referenced against current values for wall following. The calibration stages use PI (proportional, Integral) controllers while the wall following behavior is purely P (proportional).
***Sorry for any poor video quality. I have no means of editing at the moment.
[youtube]pcNQixSV63A[/youtube]
-
another great adaptation/addition to your bot. Your new sytem of wall tracking will help heaps when it comes to the hard code, not that the hard code will stay the same for very long ;D
-
another great adaptation/addition to your bot. Your new sytem of wall tracking will help heaps when it comes to the hard code
when you say hard code what do you mean? wall following with this type of drive train wasn't exactly a cake walk :D.
-
The next behavior I plan to implement is firefighting just like I did with woody.
[youtube]SgPHQ5h_wlQ[/youtube]
I already strapped a speaker to the back of GatorKS and have proven it can put out candles and such up to 6" away. The blower is intentionally weak to show off the precision and accuracy of the robot. The speaker-blower mechanism is the white thing on the back of GatorKS in the previous video.
Question what behaviors do you guys think I should program into GatorKS. At the moment my sensor array includes sharp IR on the left and right sides, sonar on the back, and a panning sonar in the front of the robot. I'm gonna put a wiimote on a tilting servo mechanism on the back of the robot very soon. I want to come up with some behaviors that show off the unique drive train, like my wall following behavior. The wiimote will enable GatorKS to track IR blobs. Input is appreciated, I want a sweet robot to show off my skills when I go to my school career fair in the fall semester.
-
sorry i was not clear.
What i ment by hard code was the code that will eventually becoome the full control code of the robot
do you know what i mean ;D
-
What i ment by hard code was the code that will eventually becoome the full control code of the robot
umm... The way I've built the robots code is that each movement has a variable setting that can be tweaked for a given motion. for instance I built in a move function where I only have to code is move(int angle,int time,bool _stop). The angle is any angle between 0 and 360, this automatically positions the wheels at this heading. All the geometry is handled in the function. "time" represents the delay in between each degree of motion the wheel servo's move. I typically select 2 ms. Last stop determines if the wheels should stop turning during positioning. I have similar functions for pivoting the robot about different wheels, and spinning. I've basically created movement libraries. Is that the kind of code you are talking about?
-
yeh thats it ;D
-
Whats that spinning thing on top of woody?
-
it looks like a wiimote, esp. since he mentioned using a wiimote to track ir
-
Its a wiimote. They connect over bluetooth and can be used for ir tracking. The remote has a resolution of 1024x768. The remote sends back the center of mass pixel value (x,y) for IR blobs. I mounted it on a stepper motor for a continuously rotating turret. I've got some other vids that may help explain. I just have to upload them to youtube first.
-
This video was taken very early on. I was demonstrating how the wiimote tracking could be performed while woody did obstacle avoidance. As you can see Woody's obstacle avoidance code was less than desirable. The robot seemed way to claustrophobic. However the wiimote tracking worked very well, which was the point of the demonstration.
[youtube]7GOyWxkQbmY[/youtube]
-
I just added the wiimote tilting mechanism to GatorKS I will be extinguishing flames soon.
-
I just got x and y axis tracking via wiimote. This is got to be my best work so far. I used PI controllers for both axis. The robot was incredibly smooth and responsive. I've actually already got the bot putting out the flame, but its kinda loud, and my parents are sleeping (I'm home for the summer).
[youtube]RjATFO7m7Fs[/youtube]
-
that is brilliant sonic ;D
-
that robot is shaping up to one hunk of killing robot!
props to you, this is all going really fast!
cant wait to see this guy with mounted lasers soon 8)
-
gr8 work sonic
-
Nicely done! Is it easier/cheaper to interface the Wii mote than a Termopile array? Does the Wii mote "see" humans also? I was considering using a TPA81 (http://www.acroname.com/robotics/parts/R255-TPA81.html) termopile array for my robot to be able to detect humans and flames, but if this is a cheaper solution, I'll gladly switch.
-
Wow . . . thats sooooo much more accurate and reliable than the flame tracking I did with the Blackfin camera . . .
-
Awesome man! ::) ::)
-
Wow . . . thats sooooo much more accurate and reliable than the flame tracking I did with the Blackfin camera . . .]Wow . . . thats sooooo much more accurate and reliable than the flame tracking I did with the Blackfin camera . . .
Thanks, Its likely due to the PI controllers. They really smooth things out.
Nicely done! Is it easier/cheaper to interface the Wii mote than a Termopile array?
probably not easier. The way I do it is a bit conveluted. but the wiimote approach will only set you back $40 if you already have xbee or bluetooth.
1) Use a program called glovepie to convert the x,y values of the remote into a PPJOY joystick (a virtual joystick).
2) I use Processing to read the joystick data and encode it into Serial data.
3) I use a bluesmirf to get the the data from Processing to my microcontroller.
Processing has some weird quirks. I discovered that it likes to wait and send serial data in packets which is craappy for this. However if you tell Processing to print the xposition some spaces, and then the yposition for some reason everything is peachy.
just in case anybody wants to know I encoded the wiimote data pretty ingeniously (I'm biased lol). As you may know ascii charters like '\r', '\n', and ' ' are less than 33. To save processing time and ease I converted the resolution of the wiimote to 100x100. I mapped the data as char datatype to send out via uart. 33-133 is an x value, 134-234 is a y value. This allows me to send spaces and newlines from Processing, and requires only two 3 statements in my micro controller code for sorting. Its a tad sneaky 8).
Does the Wii mote "see" humans also?
no, unless your sporting a couple of IR leds ;).
-
Wow . . . thats sooooo much more accurate and reliable than the flame tracking I did with the Blackfin camera . . .
@Admin or anyone else with a blackfin
I bought a blackfin too, but have been very disappointed. Mine from the start unreliably would turn on, and I haven't been able to make it communicate with my microcontroller. I can talk to a pc no problem (ftdi cable) but when I use my micro all I get is garbage data even though the baud rate is correct. did you have any problem implementing communication due to the blackfin being a 3.3V device. I turned on the pullup resistors on my mcrocontroller's RX pin to try and help. Still nothing. My other thought is that the error associated with a 115200 baud on a 16mhz (on the uC side) crystal could be causing the problem, but I communicate at that baud all the time. If you or anyone else has suggestions I'd appreciate it.
-
Thanks, Its likely due to the PI controllers. They really smooth things out.
Nah, its just that the Blackfin has trouble distinguishing the flame from the background. In the IR spectrum, assuming no sunlight, its like a bright light in a dark room.
I bought a blackfin too, but have been very disappointed. Mine from the start unreliably would turn on, and I haven't been able to make it communicate with my microcontroller. I can talk to a pc no problem (ftdi cable) but when I use my micro all I get is garbage data even though the baud rate is correct. did you have any problem implementing communication due to the blackfin being a 3.3V device. I turned on the pullup resistors on my mcrocontroller's RX pin to try and help. Still nothing. My other thought is that the error associated with a 115200 baud on a 16mhz (on the uC side) crystal could be causing the problem, but I communicate at that baud all the time. If you or anyone else has suggestions I'd appreciate it.
Do you flush the buffers? Did you use my blackfin code?
-
flush the buffers?could you elaborate?
no I havn't tried your code yet. I didn't get that far. If I can't read the version than I definitely cannot receive data. I have to fix the communication problem before I can truly communicate :P.
-
Yea definitely look at my code then. Or just use it :P
Its a single file, and my ERP is a good example of using it.
-
Yea definitely look at my code then. Or just use it Tongue
Its a single file, and my ERP is a good example of using it.
I don't follow you all the data I'm getting from the blackfin is garbage on my mcu. I'm connecting it just like you did, I looked over your tutorial. Even though I set the correct baud and I'm getting rid of flow control the mcu just recieves stuff like $@&^%%(R&^R$&#*&* instead of "Blackfin version 1" or whatever the 'V' command returns. I'm simply connecting the camera to one uart and outputting the data to another through hyper terminal. I've looked over your code and I see what needs to be done in software, but at the moment I've got a hardware problem. I'm pretty sure I have everything hooked up right because the same arrangement works with an ftdi cable to my PC.
-
Nicely done! Is it easier/cheaper to interface the Wii mote than a Termopile array?
probably not easier. The way I do it is a bit conveluted. but the wiimote approach will only set you back $40 if you already have xbee or bluetooth.
1) Use a program called glovepie to convert the x,y values of the remote into a PPJOY joystick (a virtual joystick).
2) I use Processing to read the joystick data and encode it into Serial data.
3) I use a bluesmirf to get the the data from Processing to my microcontroller.
Does the Wii mote "see" humans also?
no, unless your sporting a couple of IR leds ;).
Well, in this case a termopile array looks like it's much easier to interface (I2C comm), is able to "see" body heat and flames and looks like it's overall cheaper. I guess coupled with my AVRcam and an ultrasonic sensor, all mounted on a pan/tilt mechanism I get everything I need. Cool!
-
I just got GatorKS to chase down an put out an oil lamp. Its a little rough but I think with a little time I will get the thing running quite smoothly. The speaker used for blowing out the lamp is a little under powered, and the robot does not always position itself quite right. This is why the robot sits so long trying to blow out the flame. There isn't anything I can do about the speaker, but I should be able to tune the robots placement a bit better. Right now the robot is performing with ~%90 success rate, but some times it takes a while for the flam to finally go out.
[youtube]XVLXg9lcECQ[/youtube]
what should I make GatorKS do next?
-
make it semi autonomous like make it remote controlled when u want and the make it autonomous when required .like sort of switching modes .
-gaurav
-
Cool! You use a PWM signal to run the speaker? What power amplifier you have? And what frequency is the signal?
Thanks!
-
make it semi autonomous like make it remote controlled when u want and the make it autonomous when required .like sort of switching modes .
-gaurav
I've actually already done that lol. I used it for when I was tuning wall following ;D
Cool! You use a PWM signal to run the speaker? What power amplifier you have? And what frequency is the signal?
Thanks!
for the speaker circuit I used an sn754410 with external 1A diodes, and a not gate. The sn754410 is enabled all the time. The pwm and inverted pwm are connected to the two direction lines. This makes the speaker operate both in push and pull. I have also attached my test code. I prototyped on an atmega 168, then ported the code to my atmega640. The frequency I used was 100 Hz... unless I calculated wrong :P. Either way it works.
-
What type of speaker? Base is better?
-
What type of speaker? Base is better?
The speaker is 668-1125-ND from digikey. The speaker has a range of 100Hz to 20kHz.
If you mean "bass" as in low frequency, then the answer is yes. The lower the frequency the more deflection of the speaker.
here is a picture of the assembly to give you an idea of the mechanism. There is a .5" wood spacer to create an air chamber in front of the speaker. Then there is a .125" acrylic faceplate attached to the wood that has a sub 1/4" hole can't remember the exact diameter. This arrangement creates a focused beam of air.
**I've got more and better video too. I just have to upload it.
-
Here is another video of GatorKS putting out a flame.
[youtube]fdVDwxpHKzk[/youtube]
-
This is such a cool idea! I am thinking of a Tiny25 (has direct and inverted PWM outputs) tied to a single H-Bdridge IC (FAN8082) and a speaker. The Tiny25 also has USI that can be used as TWI. Wow!
About the air chamber. Does it have 2 faceplates, one with a hole and the other without a hole? Or both have the same hole diameter?
-
the wood is cut to the shape of the speaker and the acrylic makes an airtight fit with the exception of the hole in the center. The four holes in all three parts have 6-32 bolts holding everything together.
I'll go get a picture of it close up...
EDIT: Here it is. The hole you see is the only way for air to get in or out
-
Thanks, now I get it! Do you think if I add a hose to that hole and attach the hose to my robot's arm so he can point it to the flame to blow it off, will that slow down the air flow and need more air power?
-
Its likely the tube may unfocus the air flow. What I've found out through experimenting is that the cavity, hole size, and speaker frequency have a lot to do with performance. if you decide to make one you should play with these variables one at a time until you get an optimal setup. One things for sure though the bigger the speaker the better. Woody had a much bigger speaker and could put out fire from much further away.
-
I took a few days of respite, but now I got GatorKS putting out fire with a lot more finesse. The jitters seen in the last video are gone.
[youtube]bl3XjNDbe40[/youtube]
-
wow, great robot sonic! I would have named it delta though...
-
Why Delta ::)
-
because of the shape of the bot 8)
-
I Get your thinking