Research Papers

Design of a Clutch–Spring Knee Exoskeleton for Running

[+] Author and Article Information
Grant Elliott

Biomechatronics Group,
Department of Electrical Engineering
and Computer Science,
Massachusetts Institute of Technology,
Cambridge, MA 02139
e-mail: gelliott@alum.mit.edu

Andrew Marecki

Biomechatronics Group,
Department of Mechanical Engineering,
Massachusetts Institute of Technology,
Cambridge, MA 02139
e-mail: amarecki@alum.mit.edu

Hugh Herr

Biomechatronics Group,
MIT Media Laboratory,
Massachusetts Institute of Technology,
Cambridge, MA 02139
e-mail: hherr@media.mit.edu

Manuscript received July 17, 2013; final manuscript received June 2, 2014; published online July 21, 2014. Assoc. Editor: William K. Durfee.

J. Med. Devices 8(3), 031002 (Jul 21, 2014) (11 pages) Paper No: MED-13-1176; doi: 10.1115/1.4027841 History: Received July 17, 2013; Revised June 02, 2014

Because the leg is known to exhibit springlike behavior during the stance phase of running, several exoskeletons have attempted to place external springs in parallel with some or all of the leg during stance, but these designs have failed to permit natural kinematics during swing. To this end, a parallel-elastic exoskeleton is presented that introduces a clutch to disengage the parallel leg-spring and thereby not constrain swing-phase movements of the biological leg. A custom interference clutch with integrated planetary gear transmission, made necessary by the requirement for high holding torque but low mass, is presented and shown to withstand up to 190 N·m at 1.8 deg resolution with a mass of only 710 g. A suitable control strategy for locking the clutch at peak knee extension is also presented, where only an onboard rate gyroscope and exoskeletal joint encoder are employed as sensory inputs. Exoskeletal electromechanics, sensing, and control are shown to achieve design critieria necessary to emulate biological knee stiffness behaviors in running.

Copyright © 2014 by ASME
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Fig. 1

Torque and angle of the biological knee during unassisted human running. Dashed lines indicate least squares linear fits during stance and swing, showing the two-stiffness knee behavior in running. Data were replotted from Ref. [17].

Grahic Jump Location
Fig. 2

Parallel elastic lower limb exoskeletons. (a) Yagn's concept [14] and (b) MIT's hopping exoskeleton [15].

Grahic Jump Location
Fig. 3

Two configurations of the quasi-passive elastic exoskeleton. (a) Spanning the knee and ankle (as in Fig. 4(b)). (b) Spanning only the knee (as in Fig. 4(a)).

Grahic Jump Location
Fig. 4

Possible configurations of the clutched parallel elastic exoskeleton by combination of proximal and distal attachments

Grahic Jump Location
Fig. 5

Simulated torque applied to the knee by the spring geometry shown in Fig. 3(b) as a function of change in knee angle, starting from knee angles of 15 deg (- - -), 20 deg (—), and 25 deg (- - - - - - - -). Note that torque increases slightly sublinearly with angle and is weakly dependent on the initial knee angle when the clutch is locked.

Grahic Jump Location
Fig. 6

Schematic depiction of the designed exoskeletal joint, showing the clutch, integrated planetary gearbox, and instrumentation. The system, including the orbit of the planetary gearbox, is depicted as mechanically grounded to the proximal mount, so that flexion of the knee corresponds to rotation of distal mount. With the clutch engaged, the distal mount cannot rotate relative to the proximal mount, forcing the leaf spring to flex instead. Note that though the actual device is functionally equivalent to this depiction, its construction varies significantly.

Grahic Jump Location
Fig. 7

Renders of the left exoskeletal joint showing integration between the subsystems. (a) Lateral assembly removed to show encoder disk, solenoid, and translating mechanism. (b) Lateral planet carrier removed to show planetary gearbox and linear bearings containing bosses of the translating clutch plate. (c) Reverse angle with medial components removed to show rotating (foreground) and translating clutch plates in the disengaged state.

Grahic Jump Location
Fig. 8

State machine that controls the exoskeletal knee clutch. See Sec. 2.5 for an explanation of state exit conditions.

Grahic Jump Location
Fig. 9

Finite element analysis of the clutch teeth during 75% partial engagement at maximum expected loading. Failure does not occur as the tensile yield strength of grade 2 titanium is 275 MPa.

Grahic Jump Location
Fig. 10

Finite element analysis of the interface between a planet gear (top) and the sun gear (bottom) at maximum expected loading. Plastic deformation has just begun in the planet, as the tensile yield strength of grade 4 titanium is 480 MPa, while the grade 5 sun gear remains elastic.

Grahic Jump Location
Fig. 11

Clutch plate seen from two perspectives under an optical microscope following load testing: (a) face and (b) edge

Grahic Jump Location
Fig. 12

Planet gear at the interface to the sun gear seen from two perspectives under an optical microscope following load testing: (a) edge and (b) face

Grahic Jump Location
Fig. 13

Maximum holding torque versus mass plotted on log–log axes for several hundred commercially available interference clutches (□), particle brakes (▵), and friction (applied by air, spring, wrap-spring, electromagnet, or permanent magnet) brakes and clutches (○). For comparison, both the designed and fabricated exoskeletal joints described here are also shown (large □), plotted with masses that additionally include electronics, instrumentation, and mounting.

Grahic Jump Location
Fig. 14

Depiction of events used to identify and act on phases of the gait cycle using the state machine in Fig. 8. Data shown were recorded during treadmill running from the onboard exoskeletal joint encoder, surrogating biological knee angle during swing, and rate gyroscope, surrogating biological hip rate. Numbered circles indicate state exit transitions. More than one cycle is shown to clarify the periodic nature of the gait.

Grahic Jump Location
Fig. 15

Demonstration of solenoid latency compensation predicting peak knee extension during the swing phase. Encoder data were recorded at 800 samples per second during treadmill running. The bold region represents the window of data used in the final iteration before firing the solenoid. The firing time is denoted by a circle, while the predicted extremum is denoted by a square. The dotted line denotes data after the time the solenoid is fired. Note the correspondence between the predicted and actual times of peak knee extension.




Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In