Haptic-Assisted Target Acquisition in a Visual Point-and-Click Task for Computer Users with Motion Impairments.

Haptic assistance is the process of using force feedback to aid the operator in human-computer interaction (HCI). This may take the form of guiding the operator toward a target or assisting them in its selection. Haptic feedback has previously been investigated to assist motion-impaired computer users; however, limitations of previous 2 DOF haptic target acquisition techniques such as gravity wells and high-friction-targets have hampered progress. In this paper, two new haptic-assistive techniques are presented that utilize the 3 DOF capabilities of the Phantom Omni to produce assistance that is designed specifically for motion-impaired computer users. These include haptic cones and V-shaped funnels. To evaluate the effectiveness of the new haptic techniques, a series of point-and-click experiments were undertaken in parallel with cursor analysis to compare the levels of performance. The task required the operator to produce a predefined sentence on the Windows-On-Screen Keyboard. The results of the study prove that higher performance levels can be achieved using techniques that are less constricting than traditional assistance and without many of the drawbacks. Haptic cones produced the most significant results when compared to an unassisted interface with a mean improvement of 53 percent in the number of missed clicks and 145 percent improvement in throughput.

Applying forces to elastic network models of large biomolecules using a haptic feedback device.

Elastic network models of biomolecules have proved to be relatively good at predicting global conformational changes particularly in large systems. Software that facilitates rapid and intuitive exploration of conformational change in elastic network models of large biomolecules in response to externally applied forces would therefore be of considerable use, particularly if the forces mimic those that arise in the interaction with a functional ligand. We have developed software that enables a user to apply forces to individual atoms of an elastic network model of a biomolecule through a haptic feedback device or a mouse. With a haptic feedback device the user feels the response to the applied force whilst seeing the biomolecule deform on the screen. Prior to the interactive session normal mode analysis is performed, or pre-calculated normal mode eigenvalues and eigenvectors are loaded. For large molecules this allows the memory and number of calculations to be reduced by employing the idea of the important subspace, a relatively small space of the first M lowest frequency normal mode eigenvectors within which a large proportion of the total fluctuation occurs. Using this approach it was possible to study GroEL on a standard PC as even though only 2.3% of the total number of eigenvectors could be used, they accounted for 50% of the total fluctuation. User testing has shown that the haptic version allows for much more rapid and intuitive exploration of the molecule than the mouse version.