Research Papers

Dampace: Design of an Exoskeleton for Force-Coordination Training in Upper-Extremity Rehabilitation

[+] Author and Article Information
Arno H. A. Stienen

Research Assistant of Department of Biomechanical Engineering, University of Twente, Enschede, The Netherlands; Research Associate of Department of Physical Therapy and Human Movement Sciences, Northwestern University, Chicago, IL 60611arnostienen@gmail.com

Edsko E. G. Hekman

Department of Biomechanical Engineering, University of Twente, Enschede, The Netherlands 7500 AE

Gerdienke B. Prange

 Roessingh Research and Development, Enschede, The Netherlands 7522 AH

Michiel J. A. Jannink

Cluster Manager of  Roessingh Research and Development, Enschede, The Netherlands 7522 AH; Assistant Professor of Department of Biomechanical Engineering, University of Twente, Enschede, The Netherlands 7500 AE

Arthur M. M. Aalsma

 BAAT Medical, Hengelo, The Netherlands 7553 LZ

Frans C. T. van der Helm

Full Professor of Department of Biomechanical Engineering, University of Twente, Enschede, The Netherlands 7500 AE; Full Professor of Department of Biomechanical Engineering, Delft University of Technology, Delft, The Netherlands 2600 AA

Herman van der Kooij

Associate Professor of Department of Biomechanical Engineering, University of Twente, Enschede, The Netherlands; Associate Professor of Department of Biomechanical Engineering, Delft University of Technology, Delft, The Netherlands 7500 AE

J. Med. Devices 3(3), 031003 (Aug 31, 2009) (10 pages) doi:10.1115/1.3191727 History: Received December 07, 2008; Revised June 25, 2009; Published August 31, 2009

The Dampace exoskeleton combines functional exercises resembling activities of daily living with impairment-targeted force-coordination training. The goal of this paper is to evaluate the performance of the Dampace. In the design, the joint rotations are decoupled from the joint translations; the robot axes align themselves to the anatomical axes, overcoming some of the traditional difficulties of exoskeletons. Setup times are reduced to mere minutes and static reaction forces are kept to a minimum. The Dampace uses hydraulic disk brakes, which can resist rotations with up to 50 N m and have a torque bandwidth of 10 Hz for multisine torques of 20 N m. The brakes provide passive control over the movement; the patients’ movements can be selectively resisted, but active movement assistance is impossible and virtual environments are restricted. However, passive actuators are inherently safe and force active patient participation. In conclusion, the Dampace is well suited to offer force-coordination training with functional exercises.

Copyright © 2009 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Figure 1

Dampace: dynamic force-coordination trainer. Powered hydraulic disk brakes on the rotational axes of the shoulder and elbow can apply controlled resistance torques. Additional translating degrees of freedom at the shoulder and elbow self-align the exoskeleton axes to the anatomical axes, and allow full freedom of translation of the shoulder.

Grahic Jump Location
Figure 2

Sketch of the Dampace exoskeleton and linear guidance mechanism

Grahic Jump Location
Figure 3

Collage of Dampace components

Grahic Jump Location
Figure 4

Axes alignment in exoskeletons: (a) the effects of a single misaligned axis at the shoulder. Due to exoskeleton torque Tex, the arm and exoskeleton axes rotate an angle α. If these axes are misaligned, the human joint has to translate relative to the exoskeleton axis. If the axes are fixed, this movement creates a residual shoulder force Fsh, depending on the stiffness of the skin and bone, and an equal exoskeleton reaction force Fex; (b) translating exoskeleton axes prevent these misalignment forces. If a misalignment causes a force Fex, the exoskeleton translates until this force is gone. Torques can be applied to the limb from the rotational-stiff linkage mechanism. In 3D, the effects are the same, with adding the two other rotational axes requiring only one additional linear axis; (c) the Dampace elbow joint has two extra links, on top of which a parallelogram of cables transfer the forearm orientation to the upper arm. Translation of the joint is now independent of the rotation and vice versa, removing the requirement for the elbow alignment. At the upper arm, the rotation can be controlled and measured; a torque applied here runs through the cables and drum mechanism and is applied to the forearm without causing reaction forces.

Grahic Jump Location
Figure 5

Shoulder and elbow axes of the Dampace. The three shoulder axes run parallel to the plane of elevation, negative elevation, and axial axis in Table 1. The Dampace negative elevation is positioned at a 90 deg offset on the plane of elevation axes compared with the ISB axes. These axes do not necessarily run through the glenohumeral rotation center, but the movable, rotational-stiff linkage prevents the occurrence of shoulder reaction forces (see Fig. 4).

Grahic Jump Location
Figure 6

Disk brake as used on the exoskeleton joints, powered by a series elastic actuator (SEA) mounted on the base frame. The rotation of the motor θmot is converted by the spring with stiffness Kspr and the cylinder to a pressure in the hydraulic cable. This pressure is used to control the braking torque Tbr on the exoskeleton joint. Note that the braking torque is always in the opposite direction of the joint velocity θjnt.

Grahic Jump Location
Figure 7

Sketch of the disk brake as implemented on the negative elevation axis of the shoulder

Grahic Jump Location
Figure 8

Control loop for a single disk brake (see Fig. 6), with controllers C, physical systems H, torques T, and angles α. Subscripts denote the electromotor mot, disk brake br, and exoskeleton joint jnt. Cset are the desired interaction settings, based on the measurements of the brake torque and joint angle of all the joints. The measured brake torque Tbr is a complex function of the set brake torque (by the brake pressure) and the human interaction arm torque Tarm, and therefore difficult to use in a control loop.

Grahic Jump Location
Figure 9

Torque bandwidth for a 20 N m multisine reference signal, with the frequency response function from the reference torque Tref to the measured torque Tbr. The −3 dB gain bandwidth is 18 Hz, and the −90 deg phase bandwidth is 10 Hz. The effects of the 2 m long hydraulic cable are seen by the rapidly increasing phase delay. The transport delay in the cable was found to be 5 ms.

Grahic Jump Location
Figure 10

Weight-support mechanism. The Dampace has three weight-support mechanisms, operating independently of each other and connected to the exoskeleton linkage, the elbow and the wrist. The weight-support force Fc,b at the end of the split spring beam is independent of the spring beam angle β for all angles, because the decompositioned spring force Fsp in the z-direction (Fsp,z) is always equal to distance A times the spring stiffness k. As Fc,b=Fsp,zR1/R2, the amount of weight support can be altered by changing the spring attachment distance R1. The weight-support force on the sling Fc,s here is equal to 2Fc,b in a working volume, as defined in Table 1. The cabling beam is vertically hinged roughly above the human shoulder, which, together with the small slider underneath the cabling beam, positions the weight support exactly over the wrist and elbow.

Grahic Jump Location
Figure 11

Measured accuracy of the reconstructed fingertip position. The healthy subject was seated in front of the long side of the 600×400×300 mm3 rectangular frame and asked to trace it with his fingertip. The black stripped lines represent the frame, the dark gray lines the actual trace, and the light gray lines the shadow projections of the trace. The starting position of the fingertip was at the solid black ball, with the upper arm pointing downward and the forearm forward. The lower front and right hand corners were difficult to trace due to the arm and exoskeleton being obstructed by the trunk of the subject and the ribs of the rectangular frame. In general, the fingertip was reconstructed within 20 mm of the actual position.

Grahic Jump Location
Figure 12

(a) Resistance training setup and (b) user interface, where the table of the real-world environment (a) is recreated in the computer to allow virtual control (b)–(e). Patients need to move real objects, sometimes just sliding, or in other times, lifting it to shelves by up to the shoulder level. The movements can be made more difficult by increasing the resistance torque on the shoulder and elbow joints, which the therapist can adjust via the user interface. To guide the patient in making the movement, a virtual tunnel is created (b). When the hand moves out of the tunnel (c), all the disk brakes lock until the direction of the hand force (shown with an arrow) is again aimed toward the tunnel. The desired trajectory can be altered in direction and movement height (d), or desired vertical displacement (e). The amount of current brake force is indicated by the color and size of the four visible balls, representing the axes of the shoulder and elbow.

Grahic Jump Location
Figure 13

Integrated gaming environment connected to the Dampace torques and movements. Either isometric thoracohumeral-elevation torques or isotone rotations are mapped to the gas paddle in the racing game, and either humeroulnar isometric torques or isotone rotations to the steering wheel. Good coordination of simultaneous shoulder and elbow torques is thus required for good driving control in the game and should motivate the subjects to keep exercising.




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.

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