Would it be possible to make a high voltage relatively cheap Odrive for Induction AC motors?

Hi, I am an Odrive virgin here still waiting for delivery so pls be gentle!:slight_smile:

Anyways I would like to create a robotic arm that rivals human arm speed (not strength as well, biology is kinda unbeatable in that form factor). The problem is that you need like a minimum of 100Nm torque up until 200 rpm for actuators (motor+encoder+gears) below the weight of 1kg to be able to remotely achieve that goal. Since P=TorqueXomega, we are speaking about a peak power of 2kW or more per kilogram at the low end… or you can always move the motors far from the joints they control to make the arm light like the LIMS arm, but that makes the whole arm less precise for any realistic buget of part tolerances and money.

So as I look around the market not even 4000 USD actuators based on permanent magnet BLAC motors that Odrive supports can achieve that needed torque to weight ratio over that wide range of rpms. A TESLA induction motor for 40kg can produce like 200kW of power even at slow RPMs that is needed for starting an arm movement. So it is about 5kW power/kg and Siemens touts a motor design that can do almost twice that ratio. The best permanent magnet BLAC actuators like the designs based on the MIT cheetah actuators below 1kg can do maximum 0.5kW-1kW (with water cooling), that is ridiculously low, almost an order of magnitude lower than induction motors can do albeit at a larger scale. Even induction motors that are in the same weight class as my example can do several times more performance/weight. There are many reasons for this poor performance of permanent magnet motors and mostly have to do with the permanent magnets. They add (dead) weight and do not like high temps, not even for a short time.

Good performance to weight ratio and high torque with low RPM should be the 2 most important factors for robotic applications when it comes to actuators. Given that permanent BLAC motors are so bad in the former, why noone has made an affordable Odrive for induction motors?! I know that induction motors are not synchronous and the rotor can slip relative to the magnetic field produced, but it should be just a run off the mill PID control problem to make them as precise as the BLAC motors especially that permanent motor rotors can slip as well with dynamic torque load. Is there anything I completely miss?

Not sure I am sold on induction motors being better than permanent magnet motors, i usually see the converse. That said, it all depends on application and many factors.

ODrive can drive induction motors, so you can try it. The main limitation is that the max voltage is just over 50V, while most induction motors you find will be mains voltage or higher. However this just limits the top speed, and if you only need to use a lower speed than that you can use ODrive as-is.

Are there even any <1kg induction motors available? Especially ones that can outperform permanent magnet motors in the same size class. And anyway the real weight killer for torque is the gearbox, not the motor. Harmonic Drive is the best, but expensive, and still probably not good enough.

The LIMS approach is correct. The robot dog builders have come to the same conclusion. Not to mention biology. Get as many actuators off the arm as possible, and any that you can’t, keep them as close to the shoulder as possible to minimize moment of inertia.

Here’s my design for a 3-axis shoulder, which you’re welcome to take inspiration from https://www.rcgroups.com/forums/showthread.php?3471775-Flight-controller-for-Ornithopter-!!
It’s meant for flapping wings, so pectoral muscle motion is much higher power than the others.
The elbow will use a linear actuator on the humerus, which will add some moment of inertia, but still significantly less than placing a motor at the elbow joint.

The human manus (forearm-hand unit) really is a masterpiece of bio-engineering. The wrist/finger muscles are incredibly precise, powerful, fast, and lightweight (but note that they are still placed back toward the elbow to minimize inertia). So focus on these actuators for high performance design. Solve the others by positioning them where weight isn’t a big problem in the first place.