General “One-To-Many” (OTM) concept
The “One-to-Many” (OTM) concept allows for a single actuator, e.g. electric motor to drive multiple independently actuated and controlled degrees of freedom (DoF). Each DoF has its own dedicated reservoir for controlled energy storage and release. Each reservoir can be actively coupled, either to the actuator to store energy or to the end-effector/load to release energy. Similar to the recruitment of muscle fibers in a biological muscle, OTM system can output variable stiffness actuation by controlling the recruitment of actuated DoFs coupled to a common load. The system is also capable of augmenting power by releasing energy over a much shorter period of time than was required to store that energy. Within OTM system, the state of each actuated DoF and the state of the actuator are all mechanically decoupled. This technology may prove critical for systems with many DoF that will benefit from being compact, lightweight, efficient, modular, and capable of variable stiffness and power augmentation.
Existing examples of successfully realized OTM system include a cable driven Linear OTM utilizing linear springs [3-5] and a shaft driven Rotary OTM utilizing rotary springs [1-2]. Practical hydraulic OTM systems with fast multi-valves [6] currently under development within WPI Popovic Labs are discussed in this proposal.
OTM Definition
The general “One-to-Many” (OTM) concept allows for a single actuator, e.g. electric motor to drive multiple independently actuated and controlled degrees of freedom (DoF). Each DoF has its own dedicated reservoir for controlled energy storage and release. Each reservoir can be actively coupled, either to the actuator to store energy or to the end-effector/load to release energy. Hence, within OTM system, the state of each actuated DoF and the state of the actuator are all mechanically decoupled. Any system that encompasses this definition is considered an OTM system.
OTM Power augmentation capability
The emergent feature of any OTM system is power augmentation. The OTM system is capable of augmenting power by releasing energy over a much shorter period of time than was required to store that energy. This quite practical property of OTM systems may substantially lower power requirements on a single actuator for numerous applications.
Standard pneumatic and hydraulic systems typically consist of one motor actuating many DoF. Ordinarily, these systems have a single energy reservoir that is utilized more as a safety measure than as a mean to efficiently store and release energy of pressurized gasses or other elastic elements.
The OTM presented here is different from these and other approaches because the actuator is never directly connected to the load; instead, elastic or pressurized elements, one per each DoF, are used to store potential energy. When this energy is not being stored by the motor, it is passively stored until the energy is needed at the point of actuation. As a result, the power output at each DoF is not limited by the maximum output power of the motor itself.
From Soft Robotics Exo-Musculature to OTM Concept
The OTM concept was motivated by research on Soft-Robotics Exo-Musculature utilizing Bowden cable [10-13]. The goal was to design an at-home, comfortable, and portable orthotic and prosthetic (O&P) device built off of an Exo-Musculature that could assist patients in physical therapy routines and/or everyday tasks.
The term Exo-Musculature, introduced first in [4], refers to a soft, thin, light-weight, and compliant self-actuated garment without rigid elements or singular joints.
Traditionally, assistive or augmentative systems like orthotic or exoskeleton systems consist of rigid links and joints for the lower [21] and upper body [22] extremities. Similarly, most previous actuated systems for upper body rehabilitation use rigid exoskeletons or rigid link manipulators [23-24].
However, this traditional approach limits natural DoF and reduces comfort for the user. Further, misalignments between biological and artificial joints are inevitable [25]. Misalignments occur due to: (1) substantial skin-bone relative motion, (2) changes in volume of the limb, (3) initial imprecision when putting on the exoskeleton etc. Clearly, misalignments make exoskeletons uncomfortable and in regards to the lower extremities, can even lead to skin lesions and bone fractures due to the large forces they are subject to.
When comparing Exo-Musculatures to Exo-Skeletons, the Exo-Musculature utilizes natural anatomical structures (skeletal joints and bones) to provide support for the device and as a result, maintain natural kinematic DoF [10]. As there are no pre-specified synthetic rigid joints, an Exo-Musculature avoids misalignment problems at the joint level. Misalignments that are still present in the system are less critical and can be addressed with more advanced sensory-motor control system [13].
The more fundamental problem for mechanically actuated wearable Exo-Musculature with biologically inspired variable stiffness is need for large number of independently actuated and controlled DoF.
The force and torque produced by biological muscles can be modulated by controlling the number of fibers that are activated in parallel, a process known as recruitment [25]. This modulation of force enables optimized efficiency over a wide range of loads and contraction velocities as well as accelerations [26-28]. For the human body, there are approximately 800 skeletal muscles, each of which is composed of one hundred or more individual motor units [25]. For example, biceps brachii have about 750-800 such motor units [29-30]. Each motor unit represents a single independently actuated and controlled DoF.
Consequently, if an artificial assistive Exo-Musculature is to mimic even a small percentage of human musculature, the number of actuated DoF needs to be large. For mechanically actuated devices, conventional approaches involving one dedicated electric motor per actuated DoF results in systems that are very large, heavy, and expensive. This is clearly inappropriate for a fully mobile, wearable Exo-Musculature.
The general OTM concept solves this problem as it allows for a single actuator, e.g. electric motor to drive multiple independently actuated and controlled DoF.
The OTM biologically inspired recruitment strategy
Some of the OTM actuated DoF can be coupled to the same load, performing as muscle fibers of the same muscle. Engaging simultaneously combination of those DoF provides recruitment resulting in variable stiffness performance similar to that observed in an ordinary muscle.
Hence, OTM systems are very similar to the biological architecture and dynamical performance of “muscle fibers” and muscles.
The OTM system and SEAs
On a structural level, Series Elastic Actuators (SEA) [31] and single OTM “muscle fibers” are similar; however, the two are very different on operational and control levels. While both systems introduce an elastic element between the actuator and load, SEAs do not store energy in this element; rather, SEAs passively transfer energy from a single actuator onto respective load. The value of SEA is simplicity and escape from an impractically tight position control paradigm. This property is also shared by OTM system. The OTM system, similar to SEA, is not about tight position control. Unlike a SEA, the OTM system can modulate power output and it can utilize biologically inspired recruitment strategies in practical manner by allowing for a single actuator to drive multiple independently actuated and controlled ‘muscle fiber” DoF.
OTM architecture
On a structural level, all OTM systems consist of a single actuator coupled to DoF dedicated energy reservoirs which connect back in series with the end-effector/load, Figure 1. On an operational level, no single reservoir can be coupled to both the actuator (to store energy) and load (to release energy) at the same time. These characteristics make up the architecture of every OTM system.
As a result of this architecture, actuators are mechanically decoupled from end-effectors’ dynamics and may operate in safe and potentially close to optimal regime.
Similarly, no two DoFs share the same energy reservoir. Hence, they are all mechanically decoupled from each other. Here, it is assumed that for realistic application, two or more DoFs will be active at the same time.
Figure 1: OTM System Flowchart
Conventional pneumatic and hydraulic systems often have shared energy reservoirs. If this energy reservoir is infinitely large, its state can be considered independent of the end-effectors’ dynamics. However, for finite sized reservoirs, and in particular for very small reservoirs, the state of each reservoir depends on end-effectors’ dynamics. Hence, the end-effectors’ dynamics are not decoupled from each other. Furthermore, the actuator is not decoupled from load.
Although simple shared reservoir architectures require a smaller number of structural and sensing elements, the advantages of the OTM concept include decoupled actuator and DoFs states.
OTM systems obtain controlled energy storage and release by means of clutches, brakes, transmission mechanisms etc. for mechanical drives and valves for fluid based, i.e. pneumatic and hydraulic, drives.
The emergent feature of any OTM system is power augmentation. The OTM system is capable of augmenting power by releasing energy over a much shorter period of time than was required to store that energy. For example, imagine that a system needs to output 500 W for only 20 ms every second. Ignoring energy losses, this corresponds to at least 10 J of energy stored in an energy reservoir. During a period of 980 ms, a single actuator could store energy at rate of only 10.2 W (49 times smaller input power than output power). Gain is system that can be driven with only a tiny, light weight, inexpensive actuator and powered with a tiny electric battery.
Linear and Rotary OTM
The linear OTM system was the first physical realization of the basic OTM architecture. The system was capable of storing elastic potential energy in springs through the use of a single electric motor and a sequence of active clutch mechanisms. The prototype system had a single motor connected to three independent motor units (DoF). Each motor unit consists of two clutch mechanisms and one elastic element. This allows three possible states: Charging (mechanical energy is transferred from motor to an elastic element), Neutral (energy is stored in elastic element), and finally, Release (potential energy is transferred to load).The goal of this prototype was to realize an OTM mechanism based on fast, inexpensive, lightweight, and energy efficient clutches. The solution to this was designing latching-solenoid based clutches that could axially constrain a gear and control the rotation of a spool mounted to each gear. A latching solenoid is a type of solenoid that requires an electric pulse to open or close but does not require power to remain in either state.
Figure 2 - C1, C3, C5 are the energy storage clutches. C2, C4, C6 are the energy release clutches. Input is the motor drive wheel to charge springs. E1, E2, E3, are storage springs. E4, E5, E6 are experimental load springs.
Rotary OTM
The Rotary OTM system [1-4] was built using standard OTM architecture. The primary goal when designing this implementation was to allow for motor units (DoFs) to be compact, lightweight, energy efficient, and modular. This was realized by designing independent modules that acted as motor units.
Modules were designed to connect either to the motor or to another motor unit in series. A driveshaft running the length of the module allows for energy to be transferred through itself to another unit with minimal losses due to friction.
A planetary gear based clutch allows for the drive shaft to rotate at a constant velocity, and simultaneously allow for at least one other output at all times. This allows each motor unit to engage and disengage with the main drive shaft to transfer energy to an internal elastic element without affecting any other modules.
A few features of the Linear OTM system were improved with new Rotary OTM system. Most importantly: (1) more compact rotary spring was used instead of linear spring and (2) Rotary OTM incorporates a more refinedmeans of a controlling the energy output of the system in order to achieve actuations that could vary in speed, force, and stiffness.
Figure 3: Exploded View of Clutch 1
Figure 4: Energy Storage/Release Components
Figure 5 - A) OTM Module (One DoF) Front, B) OTM Module (One DoF) Back, C) Two OTM modules showing how they could be connected in series (left) or to a motor using a coupling (right), D) Planetary gear set used in clutch.
Hydraulic OTM
TheOTM architecture can be also realized with fluid based actuation. One example is practical hydraulicOTM systems with fast multi-valves currently under development [6].
Soft hydraulic artificial muscle array with simple sensing
The advantage of hydraulically and pneumatically actuated OTM systems is that most elements are commercially available. It is rather common for hydraulic and pneumatic systems to have a single actuator engages multiple DoF. For mechanically actuated OTM systems, the majority of components used had to be custom made.
A hydraulically actuated Exo-muscle may have several advantages over pneumatically actuated Exo-muscles. The system response times are typically much faster (sound propagates faster in water than in air), energy losses are much smaller, and for incompressible fluid forces per unit area may be much larger allowing for more compact design of system with same peak force dynamical output [32-25].
Further, The Exo-Musculature directly interfacing with human body system, consisting mostly (>60%) of water, may provide mechanically more compatible media which is often considered as an advantage [33].
The currently considered system’s actuations are those similar to conventional PAM ala McKibben muscle, dating back to 1950s where pressurizing fluid causes artificial muscle compression [34-35] and those with just opposite mechanism, such that pressurizing fluid causes artificial muscle elongation [36-37]. Later designs are similar to commercially available expandable garden hose [38]. The outer non-stretchable layer is in wrinkled state when muscle is not fully extended and it defines the maximal elongation of the muscle. Its rationale is to confine muscle stretching along axial direction. The inner layer in the form of elastic tube (e.g. latex) containing fluid returns muscle to its original state when fluid is not pressurized, Figure 6.
Figure 6 - Expandable garden hose like artificial muscle
Independent of type, i.e. McKibben vs expandable hose like, hydraulic muscle can be bi-directionally controlled by controlling the pressure or total volume of stored fluid.
Another advantage of hydraulically vs. pneumatically actuated muscle is simplified integrated sensing. Pneumatically actuated muscles typically require integrated strain gauges embedded in “skin” [17] to estimate muscle length. Hydraulically actuated muscles solves that complexities by simple resistance measurements of working fluid, in this case water. It was experimentally established that resistance of water contained in muscle is fairly linear function of muscle length and may provide a good estimate of this variable [25].
Multivalve system
Imagine hydraulic system with large number of fluid powered actuated degrees freedom (DoF), Figure 10. Each actuated DoF will necessitate at least two valves to control pressure. System on the lines of the OTM concept that has separate energy storage / power augmentation unit per DoF will require at least three valves per DoF.
However, having too many valves in the system may be problematic due to cost, size, mass, and energy efficiency.
Here, this problem is addressed by exchanging the large number of conventional valves with newly designed multivalve system based on rotary disk with embedded fast valve, Figure 11.
There exists variety of similar design solution involving selective rotary disk valves for both hydraulic [39-40] and pneumatic [41] systems. For better controllability fast standard electronic valve embedded in rotary disk is anticipated.
1-to-N multivalve consists of rotary disk with single opening embedded with fast valve (<2ms to completely close or open) and another static disc with many openings, each corresponding to one DoF. Only when rotary and static disk openings are aligned and fast valve is open fluid is allowed to flow through. Distance between rotary and static disk is much smaller compared to the diameter (~1-5 mm) of disk opening, Figure 9.
Figure 9 – Renders of proposed 1-to-N multivalve
The N-to-N multivalve, Figure 13, consists of rotary disk with single opening embedded with fast valve. Rotary disk is sandwiched with two static discs with equal number of openings, each corresponding to one of DoFs. Again, only when rotary and static disks openings are aligned and fast valve is open fluid is allowed to flow through.
Consider regime in which rotary disk is rotating with constant angular velocity corresponding to period . If the angle associated with single opening is then angle associated with space between nearest neighbor openings for densest configuration is . This defines the maximal number of openings as . If opening diameter is D and if opening center is located at distance r from the center of rotary disk then .