Table 1 Previous intelligent wheelchairs evaluated with cognitively impaired individuals

Hephaestus[11, 12]

Sonar and bumper

Semi-autonomous

Wheelchair attempts to automatically steer around obstacles, or will stop before hitting an obstacle.

Joystick

Able bodied and disabled individuals. 3 had cerebral palsy, 1 with post-polio syndrome. Unknown ages.

Participants drove through three short obstacle tasks with the wheelchair’s navigation assistance and without assistance (4 times for each scenario). Objective driving performance (time/ collisions) and subjective preference was recorded.

Wheelchair’s navigation assistance was preferred by disabled individuals over no help. Navigation assistance increased the time needed to drive through the courses and collisions still occurred with navigation assistance. Findings were limited due to the short evaluation period (1 day) and lack of complexity in the obstacle courses.

Smart Wheelchair (UK Call Centre) [9, 10]

Infrared line follower and bumper

Semi-autonomous

Wheelchair has ability to follow a line on the floor, bumpers provide anti-collision function.

Various input controls supported (e.g., switch, joystick)

Children with physical/cognitive impairments. A number of studies, including a test with 4 children who have cerebral palsy. Age 5–13.

Participants received training with the Smart Wheelchair for two 1-hour sessions per week, 8 weeks. Children progressed from single-room to school environments. Driving skills and psychosocial outcomes were measured.

3 of 4 children were able to develop 3 or more independent driving skills, and parents also reported positive changes in child’s confidence, motivation and affect. Trainers were able to decrease the assistance of wheelchair as the child showed progress in driving skills.

PALMA [13]

Sonar

Fully autonomous or semi-autonomous

Sonar sensors prevent the vehicle from hitting an object. Fully autonomous mode: PALMA navigates with no user input and has no set course. Semi-autonomous: the user has various levels of control over starting/stopping and direction of travel.

4 directions and 1 stop button. Visual (LED) and audio feedback for collisions.

Children with neuromotor disorders. Tested with 5 children with various levels of cognitive impairments. Age 3–7.

Children completed multiple 15-min driving sessions, which included driving around a room and goal oriented tasks (i.e. driving through door frames). Degree of help given by the wheelchair was lowered as a child showed proficiency. An average of 6 sessions per child.

PALMA was considered a successful training/rehabilitation tool. Its various levels of autonomy allow personalized customization to a child’s impairments. All children improved to need less assistance after several sessions.

CWA (Collaborative Wheelchair Assistant) [14]

Barcode scanner and wheel odometers (for positioning)

Semi-autonomous

Wheelchair travels along preset paths (barcodes are used to define paths in environment). User can use the joystick to avoid unexpected obstacles along those paths, and then be automatically steered back to the preset path.

Joystick

Individuals with motor/cognitive impairments. Tested with 3 individuals with cerebral palsy, and 2 individuals with traumatic brain injury individuals. Age 16–48.

Participants were trained on 6 driving tasks and then navigated through a short path with fixed obstacles. 10 path sessions for each participant, alternating between wheelchair assistance and no assistance. Collisions and joystick motion were recorded.

CWA assistance was able to help users navigate through the course with no collisions. The large variability in patient impairments showed a need for adaptable interfaces. When assistance was enabled, less joystick motion was needed and it was inferred that this relaxed the driving task.

Intelligent Wheelchair (University of Zaragoza) [15]

Planar laser and wheel odometers

Semi-autonomous

Wheelchair dynamically detects obstacles in the environment and offers to the user directions of travel that will avoid the obstacles.

Custom touch and visual interface.

Young adults with cognitive impairments. Tested with 4 students with cerebral palsy. Age 11–16.

Participants were trained to use the interface first through a computer simulation (45-60 min). Field trials consisted of driving in an uncontrolled school environment (1 session, 1 week after training). Metrics on task performance and user behavior were recorded.

Overall users were able to drive through the school environment. 6 collisions occurred that needed external intervention. Reasons for collision included obstacles at lower height of laser and system errors. The degree of cognitive impairment increased the time of driving and decreased the proficiency with the interface.

Anti-collision Skirt [16]

Low force contact sensor skirt

Semi-autonomous

Contact sensor skirt will stop the wheelchair from moving towards an object when pressure on the skirt is detected.

Joystick

Older adults with cognitive impairments. Tested with 6 older adults with mild dementia. Age 65 + .

Multiple single-subject studies, where each participant was evaluated at baseline (manual wheelchair), training (12 1-hour training sessions), and extended power wheelchair use (if deemed suitable after training). Measures of safety and mobility were taken from perception of users and external caregivers in a nursing home.

Wheelchair stopped before serious collisions occurred. False or missed collisions occurred due to gaps in the skirt, bumps on the floor, or objects above the skirt. Reception and use of the wheelchair was mixed. One adult improved mobility and well-being, another did not like its usability, slow speed, and bulky appearance. Other residents were not suitable for extended use.

CARMEN (Collaborative Autonomous Robot for Mobility Enhancement) [17]

Planar laser and wheel odometers

Semi-autonomous

Wheelchair and user share control of direction at the same time. Direction output is based on sensor readings and user input.

Joystick

Various evaluations with adults. Recently tested: 18 (mostly older) adults with physical/cognitive disabilities. Age 36–84.

Participants drove the wheelchair through a household course under two conditions: 1) standalone mode – which prevents collisions, and 2) collaborative mode – where the user and wheelchair share control. At least one run in each condition.

Not all users could complete the course in standalone mode, but all users completed it in collaborative mode. Generally collaborative mode was more efficient, unless users had high cognitive ability, in which case they may have fought the assistance that the wheelchair was giving.