Best cheapest humanoid
Best cheapest humanoid
by ginge
Mon Jan 26, 2009 12:59 am
Hi all, what is the best humanoid bot to go for that used standard servos? I want to tear whatever I buy into small pieces and retro fit it with OpenServos and control it through a customised board I am experimenting with.
So, what is the best, cheapest humanoid robot to buy (here) that I can easily take apart.
Ginge
So, what is the best, cheapest humanoid robot to buy (here) that I can easily take apart.
Ginge
http://www.headfuzz.co.uk
Robotics and Electronics hacks.
Robotics and Electronics hacks.
by ginge
Mon Jan 26, 2009 6:47 pm
Thanks guys, buy it's not really about the servos themselves, but the bipedal structure. The only stipulation I have is that I can use something resembling a "standard" servo.
I keep looking at the kondo bots, but I also keep looking at just buying the ligaments individually to make a kit.
I don't want to manufacture soemthing myself (although I could) because I want other people to reproduce my efforts. What I am working on is 99% software 1% hardware.
The end goal is to release an open source robot that is better and more open than what we have at the moment.
Do you not think it would be great if you could buy X kit/bot and install new boards, servos and source to get a kickass self balancing robot with multiple input methods (bt, usb, wifi) for the cost of a few commodity MCUs?
Oh, and I have truckloads of servos in all kinds of flavours (about 200 of the damn things), so I know what to expect there.
Cheers,
ginge (OpenServo project lead)
I keep looking at the kondo bots, but I also keep looking at just buying the ligaments individually to make a kit.
I don't want to manufacture soemthing myself (although I could) because I want other people to reproduce my efforts. What I am working on is 99% software 1% hardware.
The end goal is to release an open source robot that is better and more open than what we have at the moment.
Do you not think it would be great if you could buy X kit/bot and install new boards, servos and source to get a kickass self balancing robot with multiple input methods (bt, usb, wifi) for the cost of a few commodity MCUs?
Oh, and I have truckloads of servos in all kinds of flavours (about 200 of the damn things), so I know what to expect there.
Cheers,
ginge (OpenServo project lead)
by StuartL
Tue Jan 27, 2009 11:22 am
The closest to what you describe is the RoboNova, which uses HiTec R/C-style servos with different control circuits. AFAIK there's nothing that uses completely stock servos because traditional servos just don't have the torque, holding power etc for robotic needs.
There's a lot of work going on in the Bioloid world too. Matt and I have developed an open-source framework for the Bioloid control module (CM-5) and Matt has developed a new firmware for the servo (availability, licence etc TBC).
The Bioloid kits have three huges advantage over the RoboNovas etc:
1) They're modular and reconfigurable.
2) The servos can be reprogrammed.
3) The servos (even with stock firmware) give feedback on position, speed etc.
However the RoboNova is 'better' in that it's a simpler construction and focussed as a humanoid. It depends what you want from the kit, really, and what form of servos you wish to use for the bot.
For OpenServo (which Matt and I looked at for the replacement firmware for the Bioloid servos and discounted as unsuitable) the servo platform will dictate a lot.
It dictates what frames you can use to join the robot together, how much hardware you have to buy from (a relatively constrained range of) suppliers and the programming environments open to you.
The different areas of robotics also want different things from their humanoids. The processing power required for dynamic balancing is surprisingly low but it's enough to prevent anyone doing robosoc etc from using it as their processing power needs to go into the image processing and strategies. They just want a series of scripted motions to walk around. For us, where we're aiming for a completely dynamic walking motion, the Bioloid is perfect because we can reprogram the servos as well as the control module and yet still have a well known "industry standard" platform that I can buy spares for off-the-shelf in many countries.
There's a lot of work going on in the Bioloid world too. Matt and I have developed an open-source framework for the Bioloid control module (CM-5) and Matt has developed a new firmware for the servo (availability, licence etc TBC).
The Bioloid kits have three huges advantage over the RoboNovas etc:
1) They're modular and reconfigurable.
2) The servos can be reprogrammed.
3) The servos (even with stock firmware) give feedback on position, speed etc.
However the RoboNova is 'better' in that it's a simpler construction and focussed as a humanoid. It depends what you want from the kit, really, and what form of servos you wish to use for the bot.
For OpenServo (which Matt and I looked at for the replacement firmware for the Bioloid servos and discounted as unsuitable) the servo platform will dictate a lot.
It dictates what frames you can use to join the robot together, how much hardware you have to buy from (a relatively constrained range of) suppliers and the programming environments open to you.
The different areas of robotics also want different things from their humanoids. The processing power required for dynamic balancing is surprisingly low but it's enough to prevent anyone doing robosoc etc from using it as their processing power needs to go into the image processing and strategies. They just want a series of scripted motions to walk around. For us, where we're aiming for a completely dynamic walking motion, the Bioloid is perfect because we can reprogram the servos as well as the control module and yet still have a well known "industry standard" platform that I can buy spares for off-the-shelf in many countries.