Electromagnetic taxels

(extract from EPFL’s LMTS webpage on haptic displays and from the project final report)

We have developed a scalable and manufacturable fast haptic display with 192 taxels on an 8 mm pitch with a 10 ms refresh time per taxel. The device consists of an array of 12×16 latching electromagnetic actuators using a novel magnetic shield concept, that enables high fill factor, and eliminates cross-talk.
Key parameters of our portable EM haptic display:

  • 12 x 16 independently controlled taxels
  • 10 milliseconds refresh time per taxel (enables vibration and rapid update of static images)
  • 8 mm pitch between taxels
  • 0.8 mm vertical travel
  • 200 mN holding force (ie, easy to feel)
  • latches in both up and down states

The user explores the pattern on the haptic display using his fingertips. Our goal is not oriented to Braille reading, but to provide graphical information such as geometrical figures, maps or even artwork.

Diagram and key elements assembly of the 4×4 haptic display. a) Schematic view of a single taxel and the main actuation components. b) Photo of the 6-layer PCB containing the array of planar coils. The PCB is placed on an aluminium plate supported by four standoffs. c) Top view of the magnetic layer. It is formed by the 16 moving pot-magnets, a perimeter line of fixed pot-magnets, the top and bottom elastomer membranes (not visible) and an acrylic holder. d) A 3D printed pin interface completes the device as a final layer, and is what the user touches.


Load force vs. displacement curve for one taxel, for the two studied configurations. In the pulling configuration (blue lines (1) and (2)), the up-taxel state is dominated by the spring effective constant of the membranes. In the pushing configuration (red lines (3) and (4)), up-taxel state reflects the magnet/coil repulsion force. The colored zone around each experimental curve indicates the measured dispersion for the 16 taxels of the array.


Image of the six symbols used for the perception test. The images correspond to the device set to the pushing configuration, meaning that non-actuated taxels are in their down state, while actuated taxels are in the up state.


Zarate and H. Shea, “Using pot-magnets to enable stable and scalable electromagnetic tactile displays”, IEEE Transactions on Haptics (2016) doi: 10.1109/TOH.2016.2591951

Shape Memory Polymer taxels

Working principle of Shape Memory Polymers:

(extract from EPFL’s LMTS webpage on haptic displays and from the project final report)

Here you find a Movie showing the flexible version of our 24×32 element active skin: i) displaying different patterns, ii) device operating principle, and iii) dynamic thermal imaging showing selective taxel heating. Published in Advanced Materials Technologies 2017.

Operating principle for a single SMPs taxel. Synchronizing the global pneumatic actuation and the Joule heating switches the taxel from up to down, or vice versa.

Photo and exploded view of the 4×4 SMP haptic display


24×32 haptic display with 768 independent SMP actuators, showing the BlindPad logo. The SMP actuators were made at EPFL, the control electronics and firmware by BlindPad partner IIT, and control software by BlindPad partner Geomobile.

Pictures of each separated parts used to drive in the 32×24 SMP prototype. The hardware (actuators + pumps), the driver board and the software are shown.

Picture of the 32×24 SMP prototype with the pin interface (from IIT) and the copper plate  to speed up the refresh time.

Materials used in a Spatial Learning Skills study

Here we report an extract of the materials, methods and results used in a study with blind and visually impaired children. To read more, download the paper.

A: Child performing the tactile symbol recognition test. B: Tactile symbol recognition and enumeration in noise test. C: Memory spanning of sequences of tactile symbols test.

A: Experimental setup with a tactile display on the left side and the PC running PadDraw software on the right side. Picture shows an example of trial of the spatial memory test. B: Spatial memory test with 4×4 matrix and 4 targets. C: Shapes recognition test with a rectangle (top-left) as target and three distractors.

Possible geometrical shapes of the shapes recognition test. The first row from the top shows the three possible canonical shapes. The rows from the second to the fourth show the possible distractors. Each row from the second to the fourth shows one of the three possible levels of difficulty of distractors in decreasing order (3,2,1).

Normalized accuracy enhancement (SEM indicated) across sessions in the shapes recognition test (A) and in the spatialmemory test (B) using programmable tactile displays. Asterisks indicate a significantly larger accuracy enhancement relative to the baseline (* P<0.05)

Leo, F., Cocchi, E., & Brayda, L. (2017). The Effect of Programmable Tactile Displays on Spatial Learning Skills in Children and Adolescents of Different Visual Disability. IEEE Transactions on Neural Systems and Rehabilitation Engineering25(7), 861-872.
Download the paper PDF here