Zac Mullett Engineering Services
2D Haptic Interface


Haptics is related to our sense of touch.

The study of haptics arose from the ability of humans to physically experience, and act upon their surrounding world. This sensory ability is divided into two categories; tactile and kinaesthetic. The tactile sense gives an awareness of to stimuli on the skin surface, and is how we feel textured surfaces and small vibrations. The kinaesthetic sense gives us an awareness of body state, including position, velocity and forces supplied by muscles. Research in haptics so far has concentrated on understanding the resolutions and extents of human physical capabilities, and also on applying computers and machines in taking advantage of this fundamental sense.

Haptic interfaces are an application of haptics research. Their functionality can be broken down into two individual behaviours; measuring movements and forces of a human hand (or another body part), and creating movements and forces for the user. Haptics research has been applied in the following areas:

  • Tele-operation of machinery or robots (e.g. tele-surgery)
  • Simulation of spatial environments
  • Robot teaching
  • Many forms of user interfaces
  • Communication
  • Entertainment (e.g. the Nintendo Wii Remote)



My objective was to design a simple, cost-effective 2D haptic interface prototype that could be used to test and explore a novel approach to surface modeling and haptic presentation of 3D surfaces using only a 2D plane.

The quality of a haptic interface is based on many metrics that are measured against human sensory characteristics. Metrics considered for this project are listed below.
Free-space movement
Free-space movement metric Free space must feel free; the device must not encumber the operator. During travel in free virtual space, the user should feel no forces from the device. This is achieved by minimising inertia and statically balancing the mechanism.

Stiff barrier sensation
Stiff barrier metric Solid virtual objects must feel stiff. An important metric for a force-reflecting haptic interface is the maximum stiffness of a virtual object it can reproduce. This is not limited by the structure stiffness, but the stiffness of stable control achievable.

Vector accuracy metric
Force linearity metric
Vector accuracy and force linearity
The interface should be capable of accurately rendering a desired vector force for any position in the defined workspace. This was an important requirement given the device's intended purpose of rendering convincing 3D surface slopes. The requirement highlights the importance of accuracy in each component stage; sensing, kinematic translation, electrical supply and electromechanical conversion.


Appropriate actuator selection is critical to producing an effective haptic interface. The actuators chosen would conform to the following requirements:

  • Satisfactory torque output
  • Low rotor inertia
  • Linearity in torque delivery
  • Fast electrical response
  • Infinitely positionable
  • Direct drive (no gearing)
  • Cost-effective

These requirements typify a voice coil motor. These motors are highly back-drivable and minimise on inertia, friction, elasticity and other non-linear characteristics. The basic setup of a rotary voice coil motor is illustrated below.


Most advantageously, they can be very cost-effective when you consider there is one in every dead hard drive. After extracting a few and testing their torque characteristics I found them to be adequate for producing an interface with sufficient workspace.

Mounted hard drive voice coil motor



Mechanisms convert some form of motion or force to another form. The requirements for the interface mechanism are:

  • Provide forces in 2D at end-point
  • Rigid with zero backlash
  • Low inertia and minimal friction

A 5-bar chain is the simplest mechanism that converts torque from actuated joints into 2D translational force at the endpoint. Closed chain mechanisms also provide far greater rigidity in practice than their open counterparts. The downside to a closed chain is a drastic reduction in workspace, however in this instance the rigidity is of greater importance. Kinematics modelling is used to relate jointspace torques and workspace forces. Bar lengths are optimised to maximise workspace area while providing acceptable haptic response force for the abilities of selected motors.

Closed 5-bar mechanism and workspace

A solid model of the completed mechanism is shown below. An additional linkage was necessary to maintain the orientation of the endpoint. While the addition inevitably increased inertia and friction, these were minimised by thoughtful material and bearing selection. Fixing endpoint orientation enabled the use of a very cost-effective position sensor (discussed below).



A cost-effective solution available for planar 2D position sensing is the common optical computer mouse, with a sensing resolution within an order of magnitude of human fingertips. The mouse circuit board was extracted and mounted to the mechanism. Small nylon feet were attached below to maintain focus for the mouse and minimise friction. A finely-sanded fibrous wood was chosen as the baseboard so the mouse camera would reliably detect movement.

Nylon feet below the optical mouse circuit board      Optical mouse circuit board mounted to the mechanism



Simple PWM (Pulse Width Modulation) and a H-bridge circuit was used to control current amplitude and direction for each motor. The controller used was an existing National Instruments DAQ with PWM outputs left over from a previous project (cost-effective yet again). The modulation frequency needed to be high enough to avoid a tactile vibration feeling from the pulsing. However, high switching frequencies would result in inductive losses and non-linearities in torque response. Additional filtering circuitry and a sink resistor were added to overcome these issues.

PWM and H-bridge motor control



A haptic interface requires a control system that reads changing input conditions, processes force and kinematics equations with respect to the virtual object being modeled, then finally actuate the output to the user. This process must be rapid so the user has a believable experience of the virtual object. The complete feedback control loop is illustrated below.


LabVIEW was chosen as the programming language for all control and visualisation software in the project. It is a graphical language where ‘virtual instruments’ are wired together like an electronic circuit. This is synonymous with control system diagrams, where sensor inputs and actuator outputs are wired through control blocks. The chosen controller is directly compatible with LabVIEW. This solution was entirely cost effective because both LabVIEW and the controller were previously owned.

Graphical programming in LabVIEW

LabVIEW has a number of notable advantages over other available development platforms:

  • Rapid development
  • Visual debugging
  • Extensive documentation
  • Rugged hardware connectivity
  • 2D/3D Visualisation



The completed interface is shown below. No, it's not pretty, but prototypes often aren't.

Complete prototype 2D haptic interface

It was, however, successful in creating a haptic interaction between a user and a virtual surface. Coupled with a 3D visualisation of the modeled surface, a user can believe they are traversing a 3D surface with a 2D interface. The simple step-change slope below was an effective fundamental test.

Traversing a changing slope