Certain brightness & contrast settings are optimal for viewing medical image matter such as bones, muscle, and water. User preferences also play a key role. Old medical image view stations used to enable radiologists to adjust brightness and contrast settings on a monitor using knobs. Current workstations typically use a mouse, which is more difficult to use according to radiologists. A radiologist has suggested a 2-D haptic controller that gives various haptic icons or haptic guides towards suggested brightness & contrast settings for particular parts of medical images. For example, 2 knobs -- 1 for brightness & 1 for contrast --, or a haptic mouse with 1-D for brightness & 1-D for contrast.

swindell@cs.ubc.ca

I'm building a digital version of a turntable device so people can control the speed and direction of digital music (MP3, WAV, etc) using a controller similar to a turntable. The end result should be capable of producing scratching sounds similar to those created by a traditional scratch DJ (or better or more interesting sounds). My current system reads the velocity (which includes direction) of a spinning haptic dial and sends this information to an MP3 player once every millisecond. The MP3 player then changes the playback rate of the mp3 file to correspond with the velocity of the dial. The problem is that the digital version of scratching sounds too digital. I've not looked into the code for the MP3 player extensivly but it appears that it is not interpolating samples correctly as it needs them. I feel this may only be the surface of the problem as we have done some preliminary interpolations without much improvement. Thus, the scratching sounds like digital washing instead of "dirty" scratching.

I need someone to start looking at the code of the MP3 player (the Alsaplayer) and figure out how to make it sound better. Alsaplayer is open source and there is a decent mailing list where the developers talk about improvements for future versions. The source code has a few comments but probably not as many as I would like. It it written in C++ and uses many threads. If we can successfully make Alsaplayer sound more like an analog turntable, then we might like to try adding sound models of various surfaces to the playback so we can experiement with scratching needles on vinyl, glass on wood, pins on steel, etc...

This project would give someone excellent experience with the inside of an MP3 player, sound analysis, signal processing, interpolation and C++. Also, you'd get to be part of a very cool project that aims to provide a new level of control for DJs.

tbeamish@cs.ubc.ca

D'Groove Project 2 - Turntable Motor control

The current version of my digital turntable controller features a small haptic dial (about 3 inches in diameter). It is placed on a DC motor which is attached to an encoder. It turns out that 3 inches isn't big enough to be very usable so I'd like to increase the size of the dial to something more the size of a turntable platter - 12 inches. Actually, I'd like the device to look like a turntable would potentially like it to be installed in one.

This project involves upgrading my current haptic dial to the size of a turntable platter. In fact, it would be great if we could use an old turntable. Most turntable motors are designed to go in one direction but when we use haptics, we need to change the motor's direction rapidly. Perhaps, an actual turntable motor would not suffice. In the end, the device should look more like a turntable and still operate with enough precision to produce the right haptic feeling.

This project would give someone excellent experience in motors, amplifiers, hooking hardware into a computer and controlling it and general electronics. Also, you'd get to be part of a very cool project that aims to provide a new level of control for DJs. Did I say that already?

tbeamish@cs.ubc.ca

D'Groove Project 3 - Designing User Tests for the D'Groove Turntable

I'm aiming to have completed my implementation for my digital turntable system sometime this summer...hopefully. I'll need to start user testing the system to find out how it holds up to the rest of the controllers out there. I also need to find out some statistical information on how well people perform certain tasks while using my controller. Have they improved their performance with my controller? How can we measure this.

I need someone to help design a set of user test experiments to qualitivly and quantitivly evaluate the digital turntable controller. They should also help carry out the evaluations as well as record and analyse the data. This project would give some excellent experience with user interface design principles, user testing and statistical analysis. Also, you'd get to be part of a very cool project that aims to provide a new level of control for DJs. Third time's the charm :)

tbeamish@cs.ubc.ca

Haptics Icons Project 1 - Spatial Haptic Icons Evaluation

1. Design and perform tests for Contrast Sensitivity Function (CSF) Get an understanding of what a CSF is and how the tests are to be presented. Perform tests on several (arround 4) users gathering data from them.

2. Statistical analysis of the data gathered during the CSF tests. The CSF tests provide information as to the perceptual characteristics for a specfied task. Several statistic procedures need to be run with the results from the tests to get valid data.

3. Background / literature review on Multidimensional Scaling (MDS)

4. Design and evaluate MDS tests for spatial Haptic Icons. The MDS tests have to be carefully designed and evaluated. This involves the proper selection of haptic icons to be presented as well as a careful documentation of all procedures and decisions involved. The MDS's effectiveness depends on the design and execution of these tests. Theselection criteria will be based on experience and the results from the previous CSF tests.

5. Perform the user testing and data gathering for the MDS tests.

6. Statistical / MDS analysis of the data gathered for the tests.Having the data gathered from the tests is to be followed by statistical analysis to obtain information suitable to be processed by a MDS test.This includes data validation and outlier detection/elimination.

enriquez@cs.ubc.ca

Haptic Icons Project 2 - Extend initial Haptic Icons research on discriminability into associability

The overriding question in this unexplored new aspect of haptic interaction is, "Now that we know how to make synthetic feels - what do they mean to people? How can we design feels so it is easy for people to learn associations between feels and things, and thus use them for communcation? How can people use abstract haptic signals to convey information remotely?"

We have already started by determining the psychophysical and cognitive bases for distinction and magnitude perception of of complex synthetic haptic feels, using an MDS (Multidimensional Scaling) approach to determine perceptual clustering of complex haptic waveforms (Enriquez M.Sc. Thesis; MacLean, K., Enriquez, M., Di Lollo, V., “The Perceptual Design of Haptic Icons,” In Review).

The next steps are to:

(a) Extend this initial discriminability work, i.e. what icon design parameters individuals key on perceptually; and what design parameters result in widely spaced icon percepts. We have tried one method (MDS) and a small number of design parameters. Other quite different approaches are possible and the space needs to be extended.

(b) Address associability: what do synthetic feels mean to people? What percepts - abstract or concrete - can icons be made to intuitively represent? How can we guide learning? Again, many approaches may be taken here. This is a big and exciting new area which is thus far untouched in the haptics field.

maclean@cs.ubc.ca

In an earlier project, a mechatronics student designed and built a new kind of haptic display, incorporating force feedback into a handheld form factor.

There are two threads to continuing this project:

(a) Mechatronics: building another iteration of the handheld prototype itself. At minimum, we need to correct some problems with communication and force sensing. Beyond that, an eager student is encouraged to experiment with other means of delivering haptic feedback in the handheld format (the current prototype examines just one of many possibilities).

(b) HCI: employ the current prototype or other prototype concepts in an actual control task - for example, interacting with a room-sized display, or controlling media streams in a home.

These two subprojects can be carried out in any combination or order. For a Master's project, a student might do both; for an undergraduate project, one or the other. (a) could be done in support of (b), or alternatively, you might do (b) as the basis for a new design (a).

maclean@cs.ubc.ca

Car Knobs - Embedding haptic knobs in cars

This project is the continuation of a project I'm working on with Immersion Corp. (San Jose, CA and Montreal, QB). Immersion has begun to embed low-degree-of-freedom haptic displays in automobiles and other highly constrained environments, and has obtained US research funding to study the fundamental issues of what kind of haptic feedback is most effective in these situations. It is particularly useful to carry out this sort of research in conjunction with a company intimately knowledgable of design, manufacturing, safety and robustness issues: we need to learn the "axes" of haptic feedback which will be highly communicable, reliable and consistent in the worst-case environmental conditions.

In this M.S. or Ph.D. project, Immersion will handle the sensor and actuator design part and UBC will design and test the haptic feedback. It will probably involve spending the occasional term onsite at Immersion in San Jose or Montreal, to interact more closely with their engineers and experimental setups. There is a potential for strongly influencing Immersion's entire line of embedded products and their API; Immersion is by far the largest haptic interface company in the world today.

The details of the next phase of this project is up for discussion.

maclean@cs.ubc.ca

Modularity of Sensory Attention - Psychology experiments on the modularity / independence of human multimodal sensory perception

Proponents of haptics as a user interface medium commonly refer to "offloading vision" or "offloading hearing" - the assumption is that if we supply information about the environment or a task through touch, rather than eyes or ears, then there is more visual or auditory attention left to attend to other things.

Remarkably, we don't actually know if this is the case. Is sensory attention in one big undivided pool from which all the senses draw their resources; or is it "modular", such that visual, auditory and haptic attention all have their own resources? For the kind of offloading that we refer to above, the latter needs to be true.

This project will find out! It will involve some fairly sophisticated psychological / psychophysical experimentation; there are psychologists connected with our group who will act as resources for this.

maclean@cs.ubc.ca

Visual "change blindness" is a phenonemon (studied by Ron Rensink, here at UBC - CS) whereby people don't notice what would otherwise qualify extremely obvious changes in a scene, due to another change that happens at the same time. That is, one change rendered the observers "blind" to other major and possibly important changes.

We'd like to see if a similar phenomenon applies to the haptic sense. We've tentatively named this phenomonon (which has yet to be proven!) "haptic change numbness". We will proceed by using experiment techniques analogous to those Rensink used for change blindness.

maclean@cs.ubc.ca

Michiel Van De Panne (also of this department) has developed some cool physically-based animations: see the ski stunt simulator for an example. The only problem is, the way that you drive these animations is with a regular 2-D mouse - and most people find it quite hard.

This project is to find a better way - using haptic feedback to feel the forces involved. Initial forays have been made using a Phantom haptic display, and look promising. We need to extend the approach and validate it with user studies - and go on from there. The user studies are a bit tricky because of the learning effects involved.

maclean@cs.ubc.ca

When you drive a car down the middle of a straight road, you can turn a little to the left, or a little to the right, and still stay on the road - for a while, depending on how fast you're going. However, before long you need to apply corrective action or else you'll hit the wall: if you do this soon, the corrective action can be small, but if you wait til the last minute, you'll need to do something extreme. If you wait too long, no corrective action you apply will be enough to save you.

Michiel Van De Panne and his students are developing a method for computing this "state space" of corrective action in realtime, with driving-down-the-road as one common example of where it would be useful. We're looking at ways to convey the state space to a user (e.g. a driver, through a haptically enabled steering wheel or other haptic display) in a way that makes it easy for him/her to perceive both the risk associated with the current trajectory (without startling him into a more dangerous position), and the actions he might take to get out of it. This is not as simple as it might initially seem. For example, what if there are multiple safe options - e.g. turn slightly left, and turn sharply left, both allowing the vehicle to avoid an obstacle in the middle left? What should the haptic feedback tell the driver when there is a space of possible options?

Haptic interfaces that could be available for this problem include a Phantom, a haptically enabled steering wheel or joystick, a haptic knob, ???

maclean@cs.ubc.ca

More coming... see a past paper for ideas on this. I'd like to continue this vein of work with new interaction metaphors, new targets of interaction.

maclean@cs.ubc.ca

One of the things haptic feedback is really well suited for is providing continuous, expressive input or "handles" for computer applications that are presently hindered by the limitations of mouse and keyboard, or even gestural interfaces which don't provide the benefit of programmable resistance.

There are a lot of good candidates for this, most of them highly multimodal: e.g. computer music models; drawing and sculpting; computer animation (we're expecting some new faculty next year who are interested in collaborating on this); streaming media controllers, to name a few. If you have your own ideas for applications that need this kind of handle, I'd like to hear about it.

maclean@cs.ubc.ca