Navigation and Obstacle Avoidance Help (NOAH) for Older Adults with Cognitive Impairment: A Pilot Study
Pooja Viswanathan, James J. Little, Alan K. Mackworth, Alex Mihailidis · 2011 · The Proceedings of the 13th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2011) · doi:10.1145/2049536.2049546
Summary
This paper describes NOAH (Navigation and Obstacle Avoidance Help), an intelligent powered wheelchair system designed to provide adaptive navigation assistance to older adults with cognitive impairment in long-term care settings. An estimated 60-80% of residents in long-term care facilities have dementia, and many are excluded from powered wheelchair use because their cognitive impairments make safe operation difficult. This exclusion reduces mobility and increases dependence on caregivers. Unlike fully autonomous intelligent wheelchairs that take control away from the user, NOAH provides supportive, passive assistance — it helps the user drive safely while preserving their autonomy and sense of control. The system consists of a commercially available Pride Mobility wheelchair equipped with a 4mm Bumblebee stereo-vision camera, a laptop computer, and a Quantum Logic Controller that can enable or disable wheelchair movement. Three integrated modules operate via the Robot Operating System (ROS): a Collision Detector that generates depth maps from stereo images and stops the wheelchair when obstacles are detected within a pre-specified distance threshold; a Route Planner that uses pre-mapped environments to compute optimal routes and track the wheelchair's position; and a Prompter that uses a Partially Observable Markov Decision Process (POMDP) to model the user's behaviour and cognitive state, issuing appropriate audio navigation prompts such as "turn right," "move slightly to the left/right," or "move forward" based on whether the user appears attentive, confused, or unresponsive.
Key findings
Six participants (ages 66-97, all female except one, MMSE scores 15-25 indicating mild to moderate cognitive impairment) from a long-term care facility completed the study using a within-subjects counterbalanced A-B design with 16 total runs through a Styrofoam obstacle maze. The collision avoidance module reduced the mean number of collisions for all six participants — most dramatically for Participant 1 (from 8.0 to 1.38 collisions) and Participant 6 (from 3.13 to 0.25). The wayfinding module helped participants with short-term memory impairments: Participants 1 and 3 travelled along the optimal route with the system engaged but took longer, inefficient paths without it due to memory deficits. Participant 5, who had high disorientation, also followed the optimal route more often with the system active, though delayed prompts occasionally caused missed turns. Qualitative feedback revealed diverse responses: Participants 2 and 4 said they would use such a wheelchair as long as it functioned correctly; Participant 3 wanted it to drive faster and allow more autonomy; Participant 4 was very interested but felt she would not be allowed to use it independently. Participants 1, 4, and 5 did not feel the collision avoidance system was needed — possibly because the Styrofoam obstacles were non-threatening. Participant 5 viewed the task simply as following prompts rather than navigating, while Participant 3 learned to explore the environment using the system's guidance and found shorter paths on his own.
Relevance
This study provides rare empirical data on how older adults with actual cognitive impairment interact with intelligent wheelchair technology — most prior studies used only able-bodied participants or simulation. The NOAH system's design philosophy of supportive rather than autonomous assistance is significant: it preserves user agency while providing a safety net, which is important both for dignity and for maintaining residual cognitive and motor skills. For practitioners in assistive technology and long-term care, the findings highlight that users with cognitive impairment have highly individual needs — some benefit primarily from collision avoidance, others from wayfinding, and their engagement varies based on factors like anxiety, motivation, and understanding of the system. The use of vision-based sensing (stereo cameras) rather than expensive active sensors (laser, sonar) makes the system more portable and cost-effective. The qualitative insights — such as a participant not perceiving foam obstacles as real threats, or another viewing navigation as simply following orders — underscore the importance of ecological validity in assistive technology research and the need for longitudinal studies in real-world settings.
Tags: intelligent wheelchair · dementia · cognitive impairment · collision avoidance · wayfinding · computer vision · older adults · assistive technology · machine learning · long-term care