[JesseDarr] recently wrote in to tell us about their dynamic Arm for Robitc Mischief (dARM), a mostly 3D printed six degrees of freedom (6DOF) robotic arm that’s designed to be stronger and more capable than what we’ve seen so far from the DIY community.
The secret? Rather than using servos, dARM uses brushless DC (BLDC) motors paired with ODrive S1 controllers. He credits [James Bruton] and [Skyentific] (two names which regular Hackaday readers are likely familiar with) for introducing him to not only the ODrive controllers, but the robotics applications for BLDCs in the first place.
dARM uses eight ODrive controllers on a CAN bus, which ultimately connect up to a Raspberry Pi 4B with a RS485 CAN Hat. The controllers are connected to each other in a daisy chain using basic twisted pair wire, which simplifies the construction and maintenance of the modular arm.
As for the motors themselves, the arm uses three different types depending on where they are located, with three Eaglepower 8308 units for primary actuators, a pair of GB36-2 motors in the forearm, and finally a GM5208-24 for the gripper. Together, [JesseDarr] says the motors and gearboxes are strong enough to lift a 5 pound (2.2 kilogram) payload when extended in a horizontal position.
The project’s documentation includes assembly instructions for the printed parts, a complete Bill of Materials, and guidance on how to get the software environment setup on the Raspberry Pi. It’s not exactly a step-by-step manual, but it looks like there’s more than enough information here for anyone who’s serious about building a dARM for themselves.
If you’d like to start off by putting together something a bit easier, we’ve seen considerably less intimidating robotic arms that you might be interested in.
Of course it’s shaky, it’s all plastic
It’s not that the problem. It’s the small one sided joints.
plastic is not the problem, with proper design plastic parts can be quite rigid,
It does look a bit thin in places for an arm. Would probably shake about as much if made from aluminum.
I’m thoroughly impressed for a DIY project and how quiet it is.
“The secret? Rather than using servos, dARM uses brushless DC (BLDC) motors paired with ODrive S1 controllers.”
So instead of using an electric motor coupled with an encoder to allow the motor to be driven to an absolute position (a ‘servo’), he used an electric motor coupled with an encoder to allow the motor to be driven to an absolute position (Odrive motor + S1 encoder + controller)?
A ‘servo’ is a class of motor + controller combination, not a specific motor winding or similar. An Odrive + S1 is a servo.
While true, this is like the difference between ball bearings and bearing balls (which everyone also calls ball bearings). These days everyone calls those actuators that RC stuff uses “servos” and any other usage is an annoying academic exercise.
Funny. Everyone I know is careful to make the distinction between servos and those toy things by calling the toy ones “RC servos”. But then, everyone I know who talks about servos are industry people, so there’s that.
servo = a hobbyists packaged small dc gear motor with a potentiometer based position sensor generally constrained to an incomplete angular motion range
Servomotor = an expensive industrial motor with built in encoders
Brushless motor + odrive and encoder = a budget conscious option for hobbyists to create a basic servomotor system.
Pedantic. It is obvious from the article that the decision was to use brushless motors with “servo” encoders rather than pre-packaged hobby servos. Sheesh.
Around 1600usd just for motors and controllers… that big of a difference comapring to nema version?
Once you’re using $150 (almost US$200 once you add encoders) specialty controllers for each motor it stops really being DIY.
I feel like once you’re at 5-10x the minimum feasible price you’re just not in the same league anymore.
It’s a cool arm though, and even with specialty parts I’m sure a lot of work went into the project, it just hits different compared to someone reverse engineering an old industrial arm’s controls, or using Nema17s, basic sensors, and a few ESP32s to make an arm that can swing a camera around.
That is pretty much my take when I see these. “Yeah, that’s cool, but an arm isn’t going to be what gets that kind of investment at home.”
Add telescopic capabilities some cameras AI and make two of them so I can do my dishes and fold my laundry I’m all in 😄
I sunk around $30K into restoring my classic car.
My neighbor bought himself a $40K boat because he likes to fish.
A coworker just put money down on that new $18k laser welder cutter package that xtool announced.
different hobbyists have different budgets. Not every project has to be a shoe string budget.
I do find projects like this impressive but they are expensive and I can’t help feeling at times that it is made much simpler by just throwing money at it. I doubt the creator needs to do much or think about the actual hardware control of it, they likely just attached the motors and odrives and that handled it for them.
Another example I have seen is a hexapod using dynamixel servos. Yes it is impressive but the servos handle so much of it for you automatically that you basically don’t need to think about the motors or how they are driven, you just tell it to move to a certain position with speed and torque limits and it just does it. It also makes the project cost thousands of dollars.
Yes the projects are still impressive and if your focus isn’t the electrical, mechanical or low level control system design then why not go for the easier, more expensive option if you can afford it but I can’t help feeling that these often overpriced (for hobbyists) components just take away a lot of the skill.
With the hexapod example using dynamixels they reduce the project complexity by a lot, all you really need then is the servos and an SBC.
It’s like the “maker portfolios” for MIT applicants that show up on YouTube every year, most of them are just kids who have been given tons of money and have bought expensive components so that they can do something that looks impressive but doesn’t actually have much complexity to it due to the expensive components used.
I really want to see (or build) an arm using a low cost integrated servomotor like the DDSM400 Direct Drive Servo Motor ($25ish from Waveshare) It’s a robot wheel by default but under the rubber there are threaded screw holes on the output face plate. It’s a bit less fancy than these motors at a much easier price given the savings from the integrated controllers.