Walking Bipedal Robot (Extended)
Loading...
18,274
Loading...
Uploader Comments (dwidgit332)
see all
All Comments (40)
-
@36mauro36 that robot makes me think of the metal gear gekko
-
@SUPERSPYPIG In progress. :-)
-
is there any follow-up of this, my friend?
-
@desfergo The brackets were purchased from lynx motion (erector set), but I machined the foot mechanisms and base plate. One warning about the brackets though, they have fairly loose tolerances, they aren't very rigid by the time you get to the tip of the leg, and the repeated impacts tend to loosen screws. Use nuts with nylon insert or some loctite to keep things from wiggling loose.
Loading...
Hi, What do you use for the brain??
Nice!!
36mauro36 6 months ago
@36mauro36 The processor is a dsPIC33F series from Microchip. It's fairly low-end for embedded processing by todays standards, but I made the PCB about 5 years ago. I've used the Gumstix Overo more recently which has a lot more processing power. The Arduino isn't quite as powerful as the Gumstix, but it probably has a larger user base so it's easier to find examples of how to do things. If you're up for making a custom PCB, then the Microchip parts are a good starting point.
dwidgit332 6 months ago
@dwidgit332 Nice, I thought the microcontrollers PIC didn't have enough power for implement the ZMP...thanks for the reply, your information help me, I want to make an humanoid robot, I think a 1 meter of height..Can you help me? I have some doubts
36mauro36 6 months ago
@36mauro36 You're right, they generally don't have the power to implement ZMP. That was one of the key benefits behind this algorithm (The Foot Placement Estimator) in that the processing requirements were significantly lower, you don't need to have a foot in contact with the ground at all times, it can handle point contact, and I didn't need expensive force sensors. As far as I know though, no one has used this alg in 3D except in simulation so far.
dwidgit332 6 months ago
@dwidgit332 Which sensors use? any accelerometer or giroscope??
36mauro36 6 months ago
@36mauro36 I cheated in that respect to reduce the cost of the experiment. There are encoders on the boom so I know the position of the body, and I modified the servos so I could get feedback and therefore know the angle of each joint from the pots. Servos like the Dynamixel products are a much better choice these days. The only other sensors are contact switches on the feet. I expect a 3D robot would need a gyro and accel though. I've been using the ADIS16405 recently which is very nice.
dwidgit332 6 months ago