Simulation of Motor Impairment in Head-Controlled Pointer Fitts' Law Task
Syed Asad Rizvi, Ella Tuson, Breanna Desrochers, John Magee · 2018 · Proceedings of the 20th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '18) · doi:10.1145/3234695.3241034
Summary
This poster paper proposes simulating motor impairments in a head-controlled mouse pointer system (Camera Mouse) to address the challenge that participants with motor impairments are not always available for accessibility research or software development testing. The Camera Mouse is a computer-vision-based system that tracks head motion to move a mouse pointer, used by people with severe motor impairments who cannot use conventional input devices. The authors developed two noise-injection algorithms: C1 adds random movement of up to 50 pixels in X and Y directions with 30% probability on each pointer movement; C2 refines this by reducing noise when pointer movement is slow (below 15 pixels), adding only +/-10 pixels of noise instead of +/-50. The system was evaluated using Fitts' Law — the standard model for evaluating pointing devices that measures throughput in bits per second as a communication rate. Four participants without motor impairments performed a target-selection task under three conditions (C1, C2, and baseline Camera Mouse) using the FittsTaskTwo evaluation tool.
Key findings
The baseline condition (C3) achieved a throughput of 1.23 bits/s, consistent with previously reported results of 1.28 bits/s for non-MI users. The C1 algorithm (consistent random noise) achieved 0.747 bits/s throughput with a 31.7% error rate, which was closest to the previously reported 0.488 bits/s throughput of a participant with motor neuron disease — representing a significant reduction from baseline while not yet fully matching the MI user's performance. C2's throughput of 0.98 bits/s was higher than C1, approximately double the MI participant's rate. Interestingly, while C2's mouse traces visually resembled the MI participant's traces more closely, C1 was the more accurate simulation based on the quantitative throughput metric. This disconnect between visual similarity and measured performance highlights the complexity of accurately simulating motor impairments. The authors note that future work will apply machine learning techniques to better match both the mouse traces and communication rate of individuals with motor impairments.
Relevance
This paper addresses a practical barrier in accessibility research: the difficulty of recruiting and accessing participants with motor impairments for testing, particularly during early-stage development. The simulation approach draws an analogy to color blindness simulators, which are widely used by designers to understand the experience of color-impaired users. While the motor impairment simulation is still preliminary — the throughput gap between simulation (0.747 bits/s) and actual MI user (0.488 bits/s) indicates room for improvement — the concept has value both for preliminary testing and for raising awareness among mainstream developers about the challenges of alternative input methods. For practitioners, the paper highlights that accessibility testing should ideally involve actual users with disabilities, but simulations can serve as useful supplements, particularly for early prototyping. The use of Fitts' Law as a cross-study comparison metric demonstrates how standardized evaluation frameworks enable meaningful benchmarking of assistive input technologies.
Tags: motor impairment · simulation · head tracking · alternative input · Fitts' law · camera mouse · usability testing · ability-based design · pointing devices