Performing Locomotion Tasks in Immersive Computer Games with an Adapted Eye-Tracking Interface
Stephen Vickers, Howell Istance, Aulikki Hyrskykari · 2013 · ACM Transactions on Accessible Computing · doi:10.1145/2514856
Summary
This paper presents an approach to designing and adapting eye-gaze interaction techniques for locomotion tasks in immersive 3D games, specifically targeting young people with severe physical disabilities. The researchers developed a "transparent overlay" technique where gaze-sensitive zones on the screen trigger avatar movement commands—looking at the top moves forward, sides trigger rotation, and combinations enable diagonal movement. A key innovation is the diagnostic test that measures an individual's "interaction reliability" (R_interaction) based on fixation accuracy, completion time, and eye tracking stability. This reliability metric drives an algorithm that automatically adapts the interface by repositioning or removing gaze zones that the user struggles to access. The system was implemented as middleware called "Snap Clutch" that converts eye-tracker input into keyboard/mouse events, allowing it to work with unmodified commercial games like World of Warcraft. The research involved two groups: four young people with cerebral palsy (ages 8-17, all GMFCS Level 5) who use eye-gaze as their primary communication method, and eight young people with muscular dystrophy (ages 15-18) who can use traditional input devices but with difficulty. Twenty able-bodied participants provided baseline comparisons.
Key findings
The diagnostic test revealed significant differences between groups: CP participants achieved median reliability of 46%, MD participants 81%, and able-bodied participants 86%. Critically, the assumption that people with severe physical disabilities retain good eye control proved false for CP participants—their involuntary head movements frequently caused the eye tracker to lose tracking. After interface adaptation, all CP participants showed marked improvement in avatar control levels (measured on a 1-6 scale): one participant improved from level 1 to 5, another from level 2 to 5, and a third from level 2 to 4. The MD group performed similarly to able-bodied participants, with 88% completing all waypoints in an orienteering task. Time to first waypoint showed significant differences: CP group median 81 seconds vs. MD group 35 seconds vs. able-bodied 31 seconds. Perhaps most importantly, the research demonstrated that able-bodied participants cannot serve as representative proxies for motor-impaired users—the validity of experimental investigations using only able-bodied participants must be questioned. The study also found considerable variation within disability categories, emphasizing that people with disabilities should not be treated as homogeneous groups.
Relevance
This research has profound implications for game accessibility and assistive technology design. The adaptive interface approach—using diagnostic testing to automatically configure interaction techniques—offers a model that could be applied beyond gaming to any gaze-controlled interface. The finding that one participant (GPA-2) subsequently learned to drive her own powered wheelchair after the gaming sessions suggests that immersive games may serve as training environments for real-world mobility skills. Two participants secured funding for Tobii CEye communication devices after demonstrating their cognitive and interaction capabilities through gaming. The emotional impact was striking: parents and teachers reported engagement levels they had never seen before. For practitioners, the key takeaway is that interface adaptation must be individualized—even within the same disability category and functional classification level, users showed dramatically different gaze patterns and reliability scores. The 3D surface map visualization of reliability across screen zones provides an intuitive way to understand where a user can and cannot reliably direct their gaze, enabling informed interface customization.
Tags: eye tracking · game accessibility · cerebral palsy · muscular dystrophy · adaptive interfaces · gaze interaction · AAC
Standards referenced: ISO 9241-9