Source-linked AI summary
A Comprehensive Review of Smart Wheelchairs: Past, Present and Future
Jesse Leaman, Hung M. La
TL;DR
The paper addresses the limited availability of comprehensive, up-to-date reviews of smart-wheelchair research. It systematically reviews international studies and technological innovations, finding a broad research effort spanning interfaces, sensing, machine vision, and navigation, while emphasizing user-specific customization and acceptance.
Problem
Recent comprehensive reviews of smart-wheelchair research trends have been limited, motivating an updated state-of-the-art synthesis for power-wheelchair technology and its users.
Method
The authors conducted a systematic literature search and reviewed smart-wheelchair innovations in interfaces, sensor processing, machine vision, and navigation.
Results
The review presents international smart-wheelchair research and identifies advances in human-computer interfaces, sensing, machine vision, and navigation, including effective Kinect applications for tracking, localization, mapping, and navigation.
Takeaways & Limitations
Future smart wheelchairs should combine multimodal control, autonomous sensing and navigation, and customization, verbal feedback, and broad chair compatibility to support people with varied disabilities.
Abstract
from arXiv · showhide
A smart wheelchair (SW) is a power wheelchair (PW) to which computers, sensors, and assistive technology are attached. In the past decade, there has been little effort to provide a systematic review of SW research. This paper aims to provide a complete state-of-the-art overview of SW research trends. We expect that the information gathered in this study will enhance awareness of the status of contemporary PW as well as SW technology, and increase the functional mobility of people who use PWs. We systematically present the international SW research effort, starting with an introduction to power wheelchairs and the communities they serve. Then we discuss in detail the SW and associated technological innovations with an emphasis on the most researched areas, generating the most interest for future research and development. We conclude with our vision for the future of SW research and how to best serve people with all types of disabilities.
I. INTRODUCTION AND REVIEW METHODOLOGY
The paper addresses the lack of a comprehensive recent review of smart-wheelchair research by systematically examining international work and its technological innovations. It aims to clarify contemporary SW technology and support improved functional mobility for power-wheelchair users.
- Motivation: People with cognitive, motor, or sensory impairments may rely on power wheelchairs but can face difficulties using traditional joysticks and performing daily maneuvering tasks.Alternative controls include head, chin, sip-and-puff, and thought-control systems.
- Smart-wheelchair research: Smart wheelchairs combine a power-wheelchair or mobile-robot base with computers, sensors, and assistive technologies to support users who cannot operate a conventional power wheelchair.The review describes SWs as systems developed partly from mobile-robot technologies.
- Research gap: The review responds to little extensive and intensive review work on smart-wheelchair research trends since major reviews published in 2005.Its stated goal is a systematic and comprehensive review of international SW research since those earlier reviews.
- Review methodology: The authors searched IEEE Xplore, Google Scholar, and PubMed for English-language studies published from 2005 to 2015, screening titles, abstracts, and full texts against prespecified criteria.Conference proceedings were included only when substantially different from journal articles by the same authors.
- Review scope: The review examines human-computer interface hardware, sensor-processing algorithms, and machine-vision innovations in smart wheelchairs.The paper organizes its discussion around past power-wheelchair development, international SW research, input methods, operating modes, human factors, and future research.
- Intended contribution: The authors expect the review to increase awareness of contemporary smart-wheelchair technology and ultimately improve functional mobility and productivity for power-wheelchair users.The paper also emphasizes assistance with daily living activities.
II. PAST: THE POWER WHEELCHAIR
Power wheelchairs evolved to support mobility through specialized drive systems, batteries, controllers, and seating, while users also developed personal modifications for safety and access. These advances established the platform for smart-wheelchair research, which integrates intelligent technology into the chair.
- Power-wheelchair foundations: Power wheelchairs serve people with mobility disabilities and can also benefit individuals whose ailments cause fatigue or pain.Mass production began in the 1950s.
- Power-wheelchair foundations: Power-wheelchair chassis may include front-, rear-, center-, or all-wheel drive, with options such as stair climbing, standing, tank tracks, and all-terrain operation.These form factors illustrate the range of physical platforms available for later smart-wheelchair integration.
- Power-wheelchair foundations: The controller forms the human-machine interface, with commercial options including hand, sip-and-puff, chin, and head joysticks.These controls accommodate users who cannot operate a standard hand joystick.
- Power-wheelchair foundations: Seating systems use cushions, padded or motorized backrests, lateral supports, and adjustable footrests to improve comfort and help prevent pressure sores.Prescription restrictions and cost can leave some potential users with generic solutions that do not meet their needs.
- User modifications: Some users with quadriplegia personally modified their wheelchairs with safety equipment, rearview cameras, and assistive technology for computer input.These modifications extended the chair beyond its basic mobility functions.
- Transition to smart wheelchairs: Earlier studies achieved technological advances for daily activities, but incorporated little assistive technology to make power wheelchairs smart.Smart-wheelchair research focuses on integrating intelligent technology into the power wheelchair.
III. PRESENT: THE SMART WHEELCHAIR
Present smart-wheelchair research spans internationally produced prototypes, diverse user-specific input methods, operating technologies, and sensor-based perception. The review highlights progress alongside continuing challenges in portability, interface adaptation, and reducing continuous user commands.
- Research has shifted from power-wheelchair development toward smart wheelchairs integrating computers, sensors, and assistive technology.
- Input methods: User-specific interface adaptation remains neglected despite many projects, motivating multimodal interfaces tailored to individual characteristics.
- Input methods: Smart-wheelchair prototypes use diverse input methods, including fingertip control, head tilt, accelerometers, brain-computer interfaces, mobile devices, and deictic control.
- Sensors and perception: Smart wheelchairs fuse internal sensors such as odometers and IMUs with external sensors to detect obstacles and improve localization under uncertainty.
- Sensors and perception: Kinect-based point-cloud sensing supports target tracking, localization, mapping, navigation, and detection of hazards such as holes, stairs, and obstacles.
- Future directions: Future systems should minimize continuous commands and learn from users’ daily activities to self-operate in common situations.
V. OPERATING MODES
Smart-wheelchair operating modes range from autonomous to semi-autonomous, with the appropriate level depending on user abilities, task demands, and environmental conditions.
- Operating modes range from autonomous to semi-autonomous according to the user’s abilities and the task at hand.
- Users unable to plan or execute routes benefit most from autonomy when spending most of their time in a controlled environment.
- Users who can plan and execute routes may benefit more from systems limited to collision avoidance.
- Semi-autonomous wheelchairs may let users select distinct operating modes for different tasks.
A. Machine learning
Smart-wheelchair intelligence combines machine-learning and rule-based algorithms with multilayer control, perception, mapping, and navigation. Research also extends to companion following, collaborative movement, and robust localization across indoor and outdoor settings.
- Smart wheelchairs use specialized algorithms to make rapid decisions about heading and nearby obstacles.
- Machine learning: Examples include neural networks for obstacle detection and route reproduction, obstacle-density histograms for sensor-input fusion, fuzzy control, and rule-based approaches.
- Control architectures: Reactive control commonly forms the lowest layer of multilayer architectures, while upper layers provide deliberative reasoning and control.
- Collaborative navigation: Companion-following systems estimate a guide’s position with laser sensors and Kalman filtering, then generate a cubic-spline target path.
- Localization and mapping: A major challenge is robust localization and navigation indoors and outdoors, especially when GPS is unreliable under tree cover.
- Localization and mapping: GPS, IMU, and wheel-odometry fusion through an extended Kalman filter supports high-accuracy localization in GPS-denied environments.
- Perception and mapping: Point-cloud systems build maps, classify terrain and obstacles, and retain obstacle records after objects leave the visual field.
- Perception and mapping: Spherical cameras with improved Prewitt edge detection can detect obstacles in most indoor environments.
D. Navigational assistance
Navigational assistance research spans collision avoidance, route planning, prompting, docking, haptic guidance, and perception-based navigation. Although many operating-mode advances improve functionality, sensor limitations still constrain autonomous navigation largely to indoor environments.
- Collision detectors stop the wheelchair when an object is detected within approximately 1 meter, preventing motion toward the obstacle.
- Path planners use a global map, an initial position estimate, and visual odometry to determine routes and detect trajectory deviations and upcoming turns.
- A POMDP-based prompter estimates users’ navigation ability and responsiveness to select audio prompts, while docking algorithms identify safe docking locations.
- Haptic guidance, head-orientation tracking, and LIDAR-based transport systems support navigation through narrow spaces, crowded environments, and vehicle-access scenarios.
- Sensor capabilities currently limit autonomous navigation to indoor environments, making all-environment autonomy a major future research focus.
VI. HUMAN FACTORS IN SMART WHEELCHAIR
Human factors distinguish smart wheelchairs from ordinary mobile robots because the wheelchair is an extension of its user. Research therefore considers user abilities, trust, comfort, privacy, and ethical responsibilities in design.
- Different combinations of symptoms can require different forms of smart-wheelchair assistance and wheelchair design.
- Many people with disabilities, approximately 40% of the disabled community, find operating a standard power wheelchair difficult or impossible.
- A smart wheelchair must feel comfortable for the person using it because it functions as an extension of that individual.
- Robot performance and attributes were the largest contributors to trust in human-robot interaction, while environmental factors played a moderate role.
- Physically assistive robots raise ethical challenges involving privacy rights, including possible deactivation of video monitors during intimate procedures.
B. Learning
Learning-related smart-wheelchair research evaluates user performance, physical strain, participation barriers, and acceptance. Findings support adapting assistance to individual abilities and preferences while addressing environmental and social constraints.
- B. Learning: Wheelchair-mounted robot-manipulator experiments combined quantitative user measures with subjective evaluations to assess effectiveness, efficiency, and satisfaction.
- B. Learning: Older participants preferred appropriate robot utterances to reduce anxiety and adaptable wheeling speeds because speed preferences varied individually.
- C. Physiology: Lower fine motor abilities were associated with more collisions and longer goal-reaching times among 23 participants.
- C. Physiology: Adapting the wheelchair’s automation level to users’ fine motor abilities can improve indoor safety and efficiency.
- B. Learning: Participation can be impeded by individual impairments, inaccessible environments, social attitudes, unsuitable assistive devices, and age or confidence-related factors.
E. Commercialization
Smart-wheelchair commercialization remains limited by cost, constrained environments, platform dependence, and nontechnical barriers. Proposed uses differ between mobility assistance, skills training, and activity evaluation.
- Smart wheelchairs may serve as mobility aids, training tools, or evaluation instruments, each requiring different user-interaction behaviors.
- Only Smile Rehab’s smart wheelchair was commercially available because the technology remained expensive and autonomous navigation was limited to modified indoor environments.
- The commercially available system required users to adopt Smile Rehab’s power-wheelchair platform.
- Research systems can integrate with current wheelchair manufacturers’ systems and support sensors, guidance algorithms, obstacle avoidance, and room-to-room navigation.
- Commercialization is hindered by liability concerns, limited human-trial participation, difficult data collection, insurer reluctance to reimburse, and strict regulations.
VII. FUTURE: CO-ROBOT
Future smart wheelchairs are envisioned as safe, reliable co-robots that support independent navigation and daily activities while adapting to individual users. Key research needs include complex-environment navigation, human–wheelchair interaction, and real-time multimodal perception.
- User acceptance: Users generally accept assistive robots, but smart wheelchairs must be safe, reliable, and sufficiently autonomous to avoid failures of new features.A WPI study found experienced wheelchair users had little tolerance for failures in new features.
- Autonomous navigation: Future users could select autonomous or semi-autonomous modes and direct wheelchairs along routes indoors and outdoors with little physical or cognitive effort.Proposed capabilities include passing through doors and navigating elevators while communicating with users and maintaining individualized profiles.
- Autonomous navigation: Earlier systems often required permanently modified environments, motivating navigation algorithms that safely operate in complex settings.Such systems require online calibration and fusion of laser, camera, IMU, encoder, and GPS data.
- Human–wheelchair interaction: Reinforcement learning is proposed for human–smart-wheelchair interaction models that incorporate user and environmental sensor feedback.Suggested inputs include Emotive sensors, virtual-reality sensors, laser scanners, cameras, and GPS.
- Multimodal perception: Real-time multimodal perception should address uncertainty by combining human thought, speech, visual, and linguistic information for interaction.The paper identifies real-time perception and response to human activity as enduring challenges.
- Prototype vision: A future iChair Australia Edition combines a retractable roof with heads-up display, robotic assistance, multiple sensors, and alternative control interfaces.The illustrated platform is built on a standing-capable Levo C3 power wheelchair.
C. Smart Wheelchair with Smart Home
The paper envisions smart wheelchairs integrated with smart homes and equipped for broader independent activity. This vision combines environmental sensing, autonomous localization, multimodal control, personalization, and modular deployment across wheelchair platforms.
- Smart-home integration: Smart wheelchairs could integrate with smart homes to provide seamless control over household appliances.This extends wheelchair functionality beyond mobility into household interaction.
- Outdoor assistance: A retractable roof could provide outdoor shelter and additional nighttime safety while driving in traffic.The feature is presented as part of the envisioned outdoor smart-wheelchair system.
- Environmental perception: Stereoscopic and spherical vision, infrared laser data, machine vision, visual tracking, and gesture recognition could construct and interpret a virtual point-cloud representation of surroundings.The proposed perception stack identifies objects within the wheelchair’s environment.
- Localization: IMU, GPS, onboard data, and Bluetooth beacon data could support localization and guide the wheelchair to destinations.These information sources are proposed for navigation through public spaces.
- Governance: Liability concerns motivate treating future smart wheelchairs like registered and insured personal electric vehicles, with standardized skills certification for users.The paper proposes licensing or certification after a wheelchair skills test.
- Adaptability and deployment: The best systems are envisioned as modular, upgradable platforms that accommodate diverse disabilities through computer vision, touch, voice, and brain control.The review also emphasizes that systems should be mountable on any power wheelchair and customizable to user preferences.
- Adaptability and deployment: Future smart wheelchairs could stream and analyze sensory data in real time through cloud computing while providing verbal feedback and individualized customization.The paper frames trust, user preferences, and human factors as conditions for these solutions to improve quality of life.