Design, modelling and characterisation of a magnetic variable stiffness mechanism

Abstract

The development of VIA has gained prominence in recent years, driven by the desire for robots to work more closely with humans to improve work productivity. This class of actuators work to vary the dynamical characteristics of robots during operation, allowing them to be more pliant when executing high-speed manoeuvres while remaining stiff and firm for tasks requiring precision, helping robots satisfy the necessary safety requirements for operating in close quarters with human workers without sacrificing too much of their performance.

The sub-class of VIAs that have received the most attention thus far has been the VSA: actuators that can modulate their apparent stiffness by manipulating elastic elements within their design. Their popularity is likely due to the ability to transmit forces without excessive energetic losses (in contrast to variable dampers) and their relative ease of construction (compared to the need to interface with flywheels and other similar devices for inertial management). Among these are a small number of magnetic VSAs leverages the interactions between permanent magnets as a contactless spring, potentially eliminating an important source of fatigue failure in VSAs that comes with the use of physical springs.

Thus far, all of the magnetic VSM rely on varying the axial overlap of a rotor magnet with surrounding magnets to achieve the change in magnetic coupling strength and thus the torque and stiffness generated by the mechanism. This linear motion is achieved through the use of lead screws. As a result, the input that goes into changing the stiffness of the mechanism does not and cannot contribute to moving the equilibrium position of the mechanism.

This work looks to introduce a new magnetic variables stiffness mechanism design. The proposed design retains the internal permanent magnet rotor common to all magnetic variable stiffness mechanisms thus far, with the variation of stiffness achieved through the manipulation of an outer ring of magnets. This allows the compliance of the resulting magnetic spring system to be varied without the need to linearly insert or retract magnets, while allowing the two inputs that go into varying the stiffness of the mechanism to both also work to move the equilibrium position of the mechanism. It also avoids the need to allocate space for the axial translation of magnets in the linear displacement design. The use of identical motors and transmissions also improves operational robustness since inventory for spare parts can be reduced and simplified. In more extreme circumstances the cannibalisation of units for spares is also more broadly useful.

Mathematical models were constructed to explore various configurations of the proposed design and used to find configurations that are the most space/ volume/ mass efficient per unit of holding torque transmittable. Simulation studies were used to corroborate the findings of the mathematical models and to identify the limits of their applicability.

A proof-of-concept prototype of the most promising configuration found was designed, built, and characterised, showing that the proposed mechanism functions as expected.