Limits...
Evaluation of an intelligent wheelchair system for older adults with cognitive impairments.

How TV, Wang RH, Mihailidis A - J Neuroeng Rehabil (2013)

Bottom Line: Measurements of safety and usability were taken and compared between the two phases.However, the objective performance (time to complete course) of users navigating their environment did not improve with the IWS.This study shows the efficacy of the IWS in performing with a potential environment of use, and benefiting members of its desired user population to increase safety and lower perceived demands of powered wheelchair driving.

View Article: PubMed Central - HTML - PubMed

Affiliation: The Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto ON, Canada.

ABSTRACT

Background: Older adults are the most prevalent wheelchair users in Canada. Yet, cognitive impairments may prevent an older adult from being allowed to use a powered wheelchair due to safety and usability concerns. To address this issue, an add-on Intelligent Wheelchair System (IWS) was developed to help older adults with cognitive impairments drive a powered wheelchair safely and effectively. When attached to a powered wheelchair, the IWS adds a vision-based anti-collision feature that prevents the wheelchair from hitting obstacles and a navigation assistance feature that plays audio prompts to help users manoeuvre around obstacles.

Methods: A two stage evaluation was conducted to test the efficacy of the IWS. Stage One: Environment of Use - the IWS's anti-collision and navigation features were evaluated against objects found in a long-term care facility. Six different collision scenarios (wall, walker, cane, no object, moving and stationary person) and three different navigation scenarios (object on left, object on right, and no object) were performed. Signal detection theory was used to categorize the response of the system in each scenario. Stage Two: User Trials - single-subject research design was used to evaluate the impact of the IWS on older adults with cognitive impairment. Participants were asked to drive a powered wheelchair through a structured obstacle course in two phases: 1) with the IWS and 2) without the IWS. Measurements of safety and usability were taken and compared between the two phases. Visual analysis and phase averages were used to analyze the single-subject data.

Results: Stage One: The IWS performed correctly for all environmental anti-collision and navigation scenarios. Stage Two: Two participants completed the trials. The IWS was able to limit the number of collisions that occurred with a powered wheelchair and lower the perceived workload for driving a powered wheelchair. However, the objective performance (time to complete course) of users navigating their environment did not improve with the IWS.

Conclusions: This study shows the efficacy of the IWS in performing with a potential environment of use, and benefiting members of its desired user population to increase safety and lower perceived demands of powered wheelchair driving.

Show MeSH

Related in: MedlinePlus

Summary of software operations performed within the IWS. A) Depth image is produced by the nDepth™ FPGA; brighter regions are closer to the sensor. B) Top-down occupancy grid (created from the depth image) is used for navigation prompting; white is free space, grey is unknown, small black regions in between the grey and white space are occupied. C) Regions of high disparity are identified and bounded with white rectangles, the zone they occupy is noted. In this case an obstacle occupies the forward-right zone. D) Zones with obstacles occupying them are blocked by the joystickDCLM (Direction Control Logic Module). In this case the forward-right movement is prevented.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3750699&req=5

Figure 2: Summary of software operations performed within the IWS. A) Depth image is produced by the nDepth™ FPGA; brighter regions are closer to the sensor. B) Top-down occupancy grid (created from the depth image) is used for navigation prompting; white is free space, grey is unknown, small black regions in between the grey and white space are occupied. C) Regions of high disparity are identified and bounded with white rectangles, the zone they occupy is noted. In this case an obstacle occupies the forward-right zone. D) Zones with obstacles occupying them are blocked by the joystickDCLM (Direction Control Logic Module). In this case the forward-right movement is prevented.

Mentions: When the SBC receives a depth image from the FPGA (Figure 2A) it performs two operations. One operation is to convert the depth image into a top-down occupancy grid of the environment (Figure 2B). The algorithm for this takes the maximum disparity (i.e. closest depth) of each column in the depth image and performs ray tracing to map these points into their corresponding world location. For a detailed description of this algorithm, the reader is referred to an earlier paper [20]. The other operation is to analyze the depth image for conjoined pixels (blobs) of high disparity, and create bounding boxes around blobs greater than a minimum size threshold (Figure 2C); these blobs are assumed to be obstacles in the environment. Disparities that are bounded relate to the desired stopping distance from obstacles. For example, if the stopping distance is 700 mm, the disparities of interest include values related to 700 mm and closer. A minimum size threshold was used in order to eliminate the effects of noise (both local and random) within the depth image, while still maintaining the ability to detect small objects (for example, canes). This method of noise reduction has been described by Murray and Little [22].


Evaluation of an intelligent wheelchair system for older adults with cognitive impairments.

How TV, Wang RH, Mihailidis A - J Neuroeng Rehabil (2013)

Summary of software operations performed within the IWS. A) Depth image is produced by the nDepth™ FPGA; brighter regions are closer to the sensor. B) Top-down occupancy grid (created from the depth image) is used for navigation prompting; white is free space, grey is unknown, small black regions in between the grey and white space are occupied. C) Regions of high disparity are identified and bounded with white rectangles, the zone they occupy is noted. In this case an obstacle occupies the forward-right zone. D) Zones with obstacles occupying them are blocked by the joystickDCLM (Direction Control Logic Module). In this case the forward-right movement is prevented.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3750699&req=5

Figure 2: Summary of software operations performed within the IWS. A) Depth image is produced by the nDepth™ FPGA; brighter regions are closer to the sensor. B) Top-down occupancy grid (created from the depth image) is used for navigation prompting; white is free space, grey is unknown, small black regions in between the grey and white space are occupied. C) Regions of high disparity are identified and bounded with white rectangles, the zone they occupy is noted. In this case an obstacle occupies the forward-right zone. D) Zones with obstacles occupying them are blocked by the joystickDCLM (Direction Control Logic Module). In this case the forward-right movement is prevented.
Mentions: When the SBC receives a depth image from the FPGA (Figure 2A) it performs two operations. One operation is to convert the depth image into a top-down occupancy grid of the environment (Figure 2B). The algorithm for this takes the maximum disparity (i.e. closest depth) of each column in the depth image and performs ray tracing to map these points into their corresponding world location. For a detailed description of this algorithm, the reader is referred to an earlier paper [20]. The other operation is to analyze the depth image for conjoined pixels (blobs) of high disparity, and create bounding boxes around blobs greater than a minimum size threshold (Figure 2C); these blobs are assumed to be obstacles in the environment. Disparities that are bounded relate to the desired stopping distance from obstacles. For example, if the stopping distance is 700 mm, the disparities of interest include values related to 700 mm and closer. A minimum size threshold was used in order to eliminate the effects of noise (both local and random) within the depth image, while still maintaining the ability to detect small objects (for example, canes). This method of noise reduction has been described by Murray and Little [22].

Bottom Line: Measurements of safety and usability were taken and compared between the two phases.However, the objective performance (time to complete course) of users navigating their environment did not improve with the IWS.This study shows the efficacy of the IWS in performing with a potential environment of use, and benefiting members of its desired user population to increase safety and lower perceived demands of powered wheelchair driving.

View Article: PubMed Central - HTML - PubMed

Affiliation: The Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto ON, Canada.

ABSTRACT

Background: Older adults are the most prevalent wheelchair users in Canada. Yet, cognitive impairments may prevent an older adult from being allowed to use a powered wheelchair due to safety and usability concerns. To address this issue, an add-on Intelligent Wheelchair System (IWS) was developed to help older adults with cognitive impairments drive a powered wheelchair safely and effectively. When attached to a powered wheelchair, the IWS adds a vision-based anti-collision feature that prevents the wheelchair from hitting obstacles and a navigation assistance feature that plays audio prompts to help users manoeuvre around obstacles.

Methods: A two stage evaluation was conducted to test the efficacy of the IWS. Stage One: Environment of Use - the IWS's anti-collision and navigation features were evaluated against objects found in a long-term care facility. Six different collision scenarios (wall, walker, cane, no object, moving and stationary person) and three different navigation scenarios (object on left, object on right, and no object) were performed. Signal detection theory was used to categorize the response of the system in each scenario. Stage Two: User Trials - single-subject research design was used to evaluate the impact of the IWS on older adults with cognitive impairment. Participants were asked to drive a powered wheelchair through a structured obstacle course in two phases: 1) with the IWS and 2) without the IWS. Measurements of safety and usability were taken and compared between the two phases. Visual analysis and phase averages were used to analyze the single-subject data.

Results: Stage One: The IWS performed correctly for all environmental anti-collision and navigation scenarios. Stage Two: Two participants completed the trials. The IWS was able to limit the number of collisions that occurred with a powered wheelchair and lower the perceived workload for driving a powered wheelchair. However, the objective performance (time to complete course) of users navigating their environment did not improve with the IWS.

Conclusions: This study shows the efficacy of the IWS in performing with a potential environment of use, and benefiting members of its desired user population to increase safety and lower perceived demands of powered wheelchair driving.

Show MeSH
Related in: MedlinePlus