Robotics and Exoskeletons
Linas JuozėnasShare
Robotics & Exoskeletons
A well-chosen device can make a movement possible, support useful practice or reduce the effort of a task. Its real value is what that change makes possible for the person.
That might mean practising steps after a stroke, holding an everyday object, taking part in supported walking or making an overhead task less demanding. Each goal calls for a different design and a different way to judge success.
Understanding these differences helps us appreciate the technology’s possibilities while asking better questions about recovery, comfort, independence and access.
Standing, reaching, walking, working and conserving energy may each matter for different reasons.
The device’s assistance, the person’s own capacity and their everyday options are distinct outcomes.
Fitting, training, support, maintenance and a usable environment belong alongside the hardware.
Assistance, rehabilitation and participation
An exoskeleton is an external structure worn on the body to support or influence movement. Some contain powered joints; others use springs or other passive mechanisms. Rehabilitation robotics is broader: it also includes machines that guide a hand, support stepping or provide a controlled exercise task without being worn as a complete frame.
The same device can serve more than one purpose, but those purposes need separate assessment. Assistance may improve performance while the device is operating. Rehabilitation aims to improve functioning through a programme of care and practice. Performance assistance may reduce effort or help someone complete a specific task.
WHO describes rehabilitation in terms of functioning, individual goals and the interaction between a person and their environment.[1] That gives us a useful starting point: choose the meaningful activity first, then consider which tools might help.
Assistance during use
Can the person complete a useful task with the device? Include the help, time and effort required.
Change after training
Does a relevant improvement remain when assistance is reduced or the training device is removed, where that assessment is appropriate?
Participation in life
Does the arrangement expand what the person can choose to do at home, work or in the community?
A person can value supported standing without expecting unassisted walking. Another may prefer a wheelchair for efficient travel while using a robot for a particular exercise. Assistive technology includes many products and the services around them; combinations are often useful.[2]
The important comparison is with the person’s actual needs and alternatives. Mobility has value through freedom, comfort and participation. It does not have to resemble one particular way of moving.
What kinds of robots and exoskeletons exist?
It helps to describe a device along several dimensions: where it contacts the body, whether it supplies external power, how it transmits force and where it is intended to be used. “Robotic” alone tells us little about these features.
An orthosis supports or modifies the functioning of a body structure; a prosthesis replaces an absent body part. An exoskeleton can function as an orthotic device, but many orthoses are simple, unpowered products. WHO’s orthotic and prosthetic standards place these products within assessment, fitting and continuing services.[3]
| Design | How it interacts with the body | Typical purpose or limitation |
|---|---|---|
| Rigid wearable exoskeleton | Frames, cuffs and joints transmit forces through attachment points. | Can provide substantial guidance or support; alignment and freedom of movement require attention. |
| Soft robotic exosuit | Textiles, cables or flexible actuators apply forces to the body. | Can supplement movement with less rigid structure; soft construction does not guarantee full body-weight support. |
| Passive support device | Springs, elastic elements or mechanical structures support or redistribute loads. | Can reduce demand during a particular task without an electric motor; assistance changes with posture. |
| Powered wearable device | An actuator supplies force or joint torque under a controller. | Can assist a selected movement; power supply, timing and behaviour during faults matter. |
| End-effector rehabilitation robot | A hand, foot or other endpoint interacts with a handle, platform or attachment. | Can guide or resist a task without matching every joint of the limb. |
| Stationary gait-training system | A treadmill, moving footplates, body-weight support or robotic leg guidance may be combined. | Enables controlled stepping practice; results in the clinic need relevant functional assessment. |
Body region is only part of the description
Upper-limb systems may support the shoulder, elbow, wrist or fingers. Lower-limb systems may assist the hips, knees or ankles. A device described as “full-body” still has specific supported joints, movements and use conditions.
Some arm-support products compensate for gravity without driving the movement. Others actively guide or resist it. Naming the mechanism is more informative than assuming every rehabilitation brace contains motors or that every wearable frame is designed for everyday walking.
Power systems, sensing and control
An actuator converts energy into mechanical action. Electric motors, pressurised air, hydraulics and other mechanisms can produce force. Transmissions, cables and linkages determine how that action reaches the body. The resulting turning effect around a joint is called torque.
Powered assistance requires an energy source, often a battery in portable systems. Capacity, carried mass, speed, load and operating conditions all affect useful runtime. A specification measured under one test condition is not a guarantee for every user’s day.
Passive assistance still has to transmit force
A spring can store energy as it deforms and return some of that energy later. A support structure can also change where loads travel. Neither creates unlimited energy. Assistance during one part of a movement can introduce resistance during another, which is why the full task should be examined.
Controllers interpret signals
Joint encoders estimate position or rotation; inertial sensors detect aspects of motion; force sensors measure loads at particular locations. Some systems use muscle electrical signals, weight shifts or buttons to initiate an action. Others follow a predefined sequence with safety checks.
A controller combines these inputs with its rules or model to choose an output. It may estimate a walking phase or a likely intended movement. That estimate is not direct access to what a person is thinking. A useful controller also needs to recognise uncertainty, unexpected resistance and situations outside its intended operation.
Assistance can take different forms
Robotic control strategies include guiding a trajectory, assisting only when needed, providing resistance or creating a task-specific challenge.[4] Each changes the interaction between person and machine. More assistance can make completion easier, while the appropriate amount for practice depends on the goal and the person’s current ability.
Machine learning is one possible method of adjusting behaviour. It is not required for every effective device, and its presence does not establish that the assistance is clinically useful or well matched to a particular person.
Forces, fit and what the measurements mean
A wearable device works through physical contact. Forces applied at a cuff, strap, shoe or belt must be tolerated where they enter the body and supported elsewhere. Reducing demand at one location can change demand at another.
Human joints also move through complex paths. A rigid mechanical hinge and a biological joint do not automatically remain aligned throughout movement. Fit therefore concerns movement, pressure, comfort and control as well as whether the frame fits someone’s height.
NIOSH highlights how body dimensions and representation in design affect the fit and accessibility of occupational exoskeletons.[5] A successful demonstration on one body type should not be assumed to describe every user.
| Measurement or score | What it can tell us | What it cannot establish by itself |
|---|---|---|
| Joint position or range | Movement measured or estimated by the system, in a specified coordinate frame. | Pain-free function, tissue health or meaningful everyday improvement. |
| Force or torque at a sensor | A load at the measured attachment, handle or device joint. | The complete internal load on a tendon, joint surface or spinal structure. |
| Surface EMG | Electrical activity detected from muscles near the electrodes. | A direct measurement of muscle force, fatigue, tissue damage or future injury. |
| Steps or repetitions | How many events met the programme’s counting rule. | How much the person contributed or whether the practice improved function. |
| Metabolic energy cost | Physiological energy demand when measured with an appropriate method. | A permanent fitness gain or the same benefit in a different task. |
| Completion time or success score | Performance under the recorded conditions and assistance settings. | Independence if setup, supervision or assistance have been excluded. |
Surface EMG deserves particular care: amplitude depends on the measurement and physiological conditions, and interpreting it as muscle force or long-term adaptation can be misleading.[6] A lower signal during an assisted task may be useful evidence about that task. It is not proof that a device prevents injuries.
Make comparisons interpretable
Record the task, speed, device settings, assistance level and any support provided. If a person reaches farther because the robot now supplies more help, the result describes a different condition. Both observations may be useful, provided the change is visible.
A helpful assessment combines system data with functional performance and the user’s experience. Ask what became easier, what became harder, and whether the improvement appears in the activity the person actually wants to do.
How robotic practice can support rehabilitation
A rehabilitation robot can provide a repeatable task, support part of a limb’s weight or make movement practice possible when unaided attempts are very difficult. It can also provide a controlled environment for observing performance and adjusting the task.
These capabilities matter when they create worthwhile practice. A high repetition count is useful only in relation to the activity, effort, assistance and recovery around it. Moving a limb through a prescribed path and learning to use it in everyday life are related but different outcomes.
Make room for the person’s contribution
Depending on the goal, a therapist may encourage an attempt before assistance begins, vary the amount of help or introduce a more demanding task. Control-strategy research discusses the challenge of providing support without removing useful opportunities for active participation.[4]
That does not mean everyone should receive minimal assistance. A person may need substantial support to participate comfortably. The question is whether the chosen support serves the current purpose, and whether the task is adjusted as circumstances change.
Connect practice with meaningful activity
An illustrative arm session might link supported reaching with placing a light object in an accessible location. A gait session might later include the turning, stopping or transfer tasks that matter in the person’s setting. The clinical team decides which activities are appropriate and how they are introduced.
Feedback should help the person understand the task. A simple result or one clear adjustment can be more usable than several simultaneous displays. Game elements may make practice appealing, but enjoyment, attendance, motor change and everyday function should each be assessed for their own purpose.
Keep the wider care plan connected
Sleep, symptom management, nutrition, communication and other therapy can affect participation. Device sessions should fit the person’s overall care and available energy. A programme that looks efficient on a machine’s log may still be difficult if travel, preparation and recovery consume most of the day.
Robotic gait training after stroke
After stroke, walking difficulties can involve strength, coordination, sensation, balance, endurance and attention in different combinations. A gait-training device may help organise stepping practice, but no single mechanism describes everyone’s needs.
What the 2025 evidence review found
A Cochrane review included 101 studies and 4,224 adults, with searches to December 2023. Electromechanical or robotic gait training combined with physiotherapy probably increased the chance of independent walking at the end of treatment. It did not establish a corresponding improvement in average walking speed or six-minute walking distance, and later follow-up did not clearly show a sustained advantage in independent walking. These findings concern varied devices and programmes.[7]
Independent walking, speed and endurance are distinct achievements. A person can improve in one while still needing help with another. The result also does not mean that every study participant used a portable exoskeleton in daily life.
A separate question: immediate assistance
In a small 2020 study, six people with chronic stroke walked faster and farther during tests with an assisting soft robotic exosuit.[8] This demonstrates a potential device-on benefit under the tested conditions. It does not establish that the same improvement remains after the assistance is removed.
Use the evidence to choose a goal
For someone practising early steps, the relevant question may be whether the programme supports meaningful progress toward walking with less human help. For someone already walking, the priority could be comfort, distance or navigating a particular setting.
The assessment should state which walking aid, orthosis, device and human support were used. “Independent” in a study is an operational definition, not necessarily a claim that the person can manage every surface, staircase or community task.
Robotic support for arm and hand function
Upper-limb robots may support the arm against gravity, guide reaching, provide resistance or assist finger opening and closing. A complete everyday action can also require sensation, object recognition, grip adjustment and coordination with the other hand.
What a large trial adds to the picture
The 2019 RATULS trial randomised 770 people with moderate or severe arm limitation after stroke. It compared a robotic programme with enhanced therapy based on repetitive functional tasks and with usual care. The robotic programme did not show a significant advantage on the primary arm-function outcome over either comparator.[9]
This finding applies to the programme studied. It is a reason to ask for evidence about a proposed intervention and its functional outcome, rather than assuming that precise movement guidance or extra technology necessarily produces better recovery.
Assisted grasp is also a useful research goal
A 2023 case series examined a soft robotic glove intended to help people with chronic stroke perform hand tasks. The study reported improvements in assisted hand performance and explored usability.[10] Such early work can identify possibilities and design problems; it does not establish lasting recovery without the glove.
For a practical trial, it helps to choose an actual object and activity: holding an item steady, releasing it at the right moment or completing part of dressing. The device must allow a suitable grasp and a dependable way to release, with attention to skin contact and sensation.
Measure the activity the person values
More range at the elbow may matter, but it is not interchangeable with managing a meal or using a phone. A functional test, the help required and the person’s own priorities give the movement measurements context.
Some useful outcomes involve ongoing assistance. Others concern capacity developed through therapy. Keeping the purpose explicit allows both to be valued honestly.
Exoskeletons after spinal cord injury
For some people with spinal cord injury, a powered walking exoskeleton can make supported standing and stepping possible. Systems differ in how they support balance, initiate steps and require assistance. Clinical and device-specific assessment is essential before use.[11]
A motor moving the legs does not by itself demonstrate repair of the spinal cord. The important questions include what the person can do with the device, what help is required and whether the experience serves a goal they have chosen.
Community use has been tested
A 2024 randomised trial involving 161 veterans with chronic spinal cord injury compared wheelchair use alone with wheelchair use plus access to a walking exoskeleton for four months at home and in the community. The exoskeleton group did not have significantly better results on the selected patient-reported mental and physical health outcomes. Device use was low, with companion availability an important barrier.[12]
This helps explain why a successful walking demonstration and a useful daily-life arrangement are different questions. The trial does not determine what every individual will value, and it does not establish that a different device or support model would have the same result.
Include the whole mobility journey
Consider getting into the device, reaching the activity, turning, stopping, resting and returning. What happens at a narrow doorway? Where can someone sit? Who can help if a problem occurs? How does the device travel between locations?
A wheelchair, adapted transport, an accessible environment and a walking device may serve different parts of the same person’s life. Time, fatigue and assistance needs can be as important as the number of steps recorded.
Claims about preventing pressure injuries, restoring bone density or reducing other complications need evidence for those specific outcomes. They should not be assumed simply because a device enables standing or walking.
Occupational exoskeletons and physical workload
Occupational devices may support the arms during overhead work, assist the trunk during bending or influence lower-limb effort. A device can make a selected posture or movement feel easier while changing how the rest of the task is performed.
NIOSH describes promising workload reductions alongside practical concerns such as fit, mobility and implementation.[13] The essential question is whether the device improves the whole task under realistic conditions, including transitions and activities it was not primarily designed to assist.
A lower workload measure is an intermediate result
Reduced muscle activity during a short experiment does not establish fewer injuries over months or years. EU-OSHA highlights uncertainty about long-term effects and the need to consider risks introduced by the device itself.[14]
For example, an arm-support device might help with a sustained overhead task but interfere with reaching a low shelf, moving through a confined area or using protective equipment. Those transitions belong in a trial, even if the main task looks successful.
Improve the work system first
The hierarchy of controls prioritises removing or reducing hazards at their source.[15] Relevant options may include changing object weight, handling height, access, equipment or the production process. Work organisation and recovery opportunities also deserve attention.
A workplace trial can then ask whether an exoskeleton adds a useful benefit for the remaining task. Include workers in defining success and in reporting discomfort or interference. Productivity alone is an incomplete measure when the intended purpose is reducing physical burden.
An easier task should not automatically lead to heavier loads, faster quotas or longer exposure. Reassess the actual workload if expectations change. Workers need a practical way to pause, report problems and request a different arrangement.
Sport, fitness and support with ageing
Powered assistance can reduce the effort needed for a movement. In a 2022 study, researchers personalised ankle-exoskeleton assistance using wearable measurements and demonstrated reduced walking energy cost in the tested conditions.[16] That is an engineering achievement with potential practical value. It does not show that wearing the device produces a lasting increase in the user’s unaided fitness.
| Goal | What to assess | A useful distinction |
|---|---|---|
| Walk with less effort | Energy cost or perceived effort, comfort and actual distance under comparable conditions. | Conserving energy during use differs from developing endurance without assistance. |
| Recover a sport-related capacity | Relevant strength, movement, tolerance and sport-specific function within a rehabilitation plan. | A machine’s range-of-motion score is one component of recovery. |
| Take part in more activity | Activities actually completed, enjoyment, fatigue and support needed. | Increased participation can matter even if unaided capacity is unchanged. |
| Maintain independence with ageing | Meaningful daily tasks, usability, falls-related assessment and other support needs. | Age alone does not establish suitability or prove a device prevents falls. |
Sports rehabilitation needs relevant progression
A robotic device may provide a controlled task or a repeatable measurement after injury. Its added value should be assessed within the rehabilitation programme. For example, the Aspetar guideline for rehabilitation after anterior cruciate ligament reconstruction addresses exercise and progression toward functional and sporting demands.[17]
That process includes more than symmetrical movement on a machine. Tissue recovery, strength, tolerance, coordination and the demands of the sport remain relevant. A device should not automatically accelerate a return date or substitute for an appropriate assessment.
Older adults have varied goals and capacities
Some people may value help during walking or standing; others need better access, suitable footwear, a walking aid or a different exercise arrangement. WHO physical activity guidance supports appropriately adapted aerobic, strengthening and balance activity in older age.[18]
A wearable assistant should be considered in relation to that person’s needs and options. Claims of fewer falls, longer independent living or prevention of frailty require direct evidence. Comfortable participation and preserving energy for valued activities are worthwhile outcomes in their own right.
Selection, fit and safe use
Suitability depends on the model, intended use and person. A device used with a harness and clinical team may have a very different role from a wearable intended for home or workplace use.
Human-factors guidance emphasises intended users, tasks and environments.[19] For powered walking exoskeletons, regulatory controls also address matters such as testing, training, appropriate users and warnings.[20] These principles help frame the questions below; they are not a universal eligibility checklist.
| Area | Questions to resolve | Why it matters |
|---|---|---|
| Physical fit | Do dimensions, joint alignment, attachments and clothing work throughout the intended movements? | A static fit can change when reaching, sitting or turning. |
| Health and functional needs | Are weight-bearing, required movements and the expected effort appropriate for this person? | The ability to fit inside a device does not establish readiness to use it. |
| Skin and sensation | Can contact areas be checked and pressure or rubbing detected and addressed? | Comfort reports alone may miss a problem when sensation is reduced. |
| Support and control | Who assists with transfers, balance, setup and stopping? Can the user operate the controls? | Support requirements are part of the actual activity. |
| Environment | Are surfaces, doorways, seating, access and any other equipment compatible with use? | A clinic demonstration may not represent the intended setting. |
| Failure and maintenance | What happens during a fault, loss of power or unexpected resistance? Who provides servicing? | A usable stop and recovery procedure must be understood before a problem occurs. |
Clinical risks need individual assessment
For people with spinal cord injury, relevant concerns can include fragile bones, skin injury, restricted movement, spasticity and difficulty tolerating upright posture. Some are also susceptible to autonomic dysreflexia. The clinical team should assess these issues and plan monitoring appropriate to the person and device.[11]
Do not assume that more weight-bearing is automatically appropriate or that the absence of pain proves safe loading. New pain, skin damage, marked dizziness or unexpected device behaviour should prompt a stop and assessment through the agreed support plan.
Learn the practical skills
Training includes putting the device on and taking it off, starting, stopping, changing position and responding to problems. Use the required supervision and environment while learning. Avoid improvising attachments or bypassing limits to make a device fit a different task.
Review changes in fit, health, surroundings or software before relying on earlier assessments. The aim is to make use understandable and dependable, with enough support to respond when circumstances change.
Cost, ethics, regulation and personal data
Look at the full cost of useful access
The relevant commitment can include assessment, fitting, training, staff or companion time, transport, maintenance, consumables and repairs. A device that is available for purchase may still be impractical if those services are missing.
WHO identifies access to assistive technology as a matter of products and the systems around them.[2] For a household or clinic, it is sensible to compare the whole arrangement with alternatives that could meet the same goal. Shared access or a supervised trial may be more useful than ownership, depending on the circumstances.
Funding and reimbursement depend on the country, provider, device and indication. Obtain current written information for the actual proposal, including what happens when a component fails or a person’s needs change.
Regulatory conformity answers a specific question
In the EU, medical devices fall within the applicable Medical Device Regulation framework, with requirements linked to their intended purpose. CE marking indicates conformity with applicable requirements; it does not establish that a device is the best option for every user or superior to another rehabilitation approach.[21]
Ask for the intended use, instructions and evidence relevant to the proposed application. A research prototype, an occupational support device and a medical rehabilitation system should not be treated as interchangeable.
Collect information for a clear purpose
Movement logs, performance records and linked health information may be personal data. Where the GDPR applies, processing needs an appropriate lawful basis, purpose limitation, data minimisation and suitable safeguards. Health data and certain identifying biometric data require additional protection.[22]
Explain who sees the records and whether they affect care, employment or access to a service. The person should be able to ask how a conclusion was reached and correct inaccurate context. A motion score should not become an unexplained judgment about effort or commitment.
EDPB guidance highlights how power imbalances can undermine freely given consent, particularly at work.[23] Participation in a device trial and permission for secondary data uses should be considered separately. Useful technology should increase a person’s options and respect a decision to pursue another approach.
A five-step framework for evaluating a device
This is an original practical framework for discussing a device with a rehabilitation team, occupational specialist or service provider. It helps turn a broad promise into a specific trial; it is not a treatment protocol.
-
Choose an activity that matters.
Describe the goal in everyday terms: practise standing, place an object, move between two locations or complete a particular work task with less burden. Clarify whether the aim is assistance during use, capacity developed through training, or participation made possible by a new arrangement.
-
Compare suitable options.
Discuss the device alongside relevant therapy, orthoses, mobility aids, task changes or environmental adaptations. Check the intended use, evidence, support requirements and likely total cost. Establish the person’s starting performance under clearly described conditions.
-
Try the complete arrangement.
Use appropriate assessment, fitting and supervision. Include preparation, transfers, controls, task transitions, rest and finishing the session. Find out what assistance is needed and whether that assistance can actually be provided in the intended setting.
-
Measure the chosen benefit and any burden.
Record the outcome, device settings, practice completed, comfort, fatigue and time required. Use comparable conditions when tracking change. If retained improvement is the goal, arrange suitable assessment beyond the assisted task; if ongoing assistance is the goal, assess real use with that assistance.
-
Review and plan continuing support.
Decide whether the benefit justifies the effort and cost for this person. Continue, adapt or choose another approach. Agree on maintenance, follow-up, help with problems and a backup arrangement. Revisit the decision when needs, equipment or circumstances change.
An example: making a household task easier
Imagine someone with limited hand function who wants to hold a light object steady while using the other hand. A therapist and the person first examine the task, positioning and available aids. They consider whether a robotic glove adds useful assistance.
A supervised trial includes putting on the glove, operating its controls, grasping and releasing the object, and checking contact areas. The team records whether the person can complete the task, how much help it takes and whether setup makes the activity impractical.
A successful result might be completing the activity more comfortably with the glove. If rehabilitation is also a goal, improvement without its assistance is assessed separately. Either way, the person’s experience helps determine whether the tool belongs in their daily routine.
Chosen activity · device and settings · help required · result · comfort and fatigue · preparation time · person’s preference · next decision.
Where robotics and exoskeletons may develop next
Better devices may emerge through improvements in fit, force transmission, control, energy use and services. Progress in one component still needs evaluation in the complete system. A lighter actuator, for example, is valuable only if the resulting device remains useful, durable and manageable.
Soft systems can address specific movement problems
A 2024 proof-of-concept study used soft robotic apparel to assist hip movement in one person with Parkinson’s disease and avert freezing of gait in several tested conditions.[24] This is a meaningful example of targeting a particular problem. One participant’s result does not establish effectiveness for the wider population.
Brain–machine interfaces expand control possibilities
In a 2019 demonstration, a participant with tetraplegia used an implanted brain–machine interface to control an exoskeleton, including walking while suspended for support.[25] The work showed a control possibility. It did not demonstrate independent walking around the community.
Neural interfaces bring their own questions about signals, training, reliability, clinical suitability and long-term support. They should be evaluated as part of the whole intervention, with clear outcomes and participant involvement.
Adaptive control should remain understandable
AI methods may help estimate movement or personalise assistance. NIST’s AI Risk Management Framework emphasises validity, reliability and context of use.[26] For a wearable robot, practical questions include how the controller behaves outside familiar conditions, whether limits remain effective and how changes are reviewed.
A more complex algorithm is useful when it produces a better interaction. Useful research measures should include relevant activity, comfort, adverse effects, time, support needs and continued use, alongside technical performance.
Design with the people who will live with it
Users can reveal priorities that a laboratory task misses: quiet operation, manageable clothing, reliable release, easier transport, repair access or less dependence on another person. These are central design requirements when the goal is everyday participation.
The future should offer more workable choices. Some will be sophisticated wearable robots; others will be simpler devices, improved services or more accessible environments. Each deserves attention when it helps someone do something they value.
Common questions
Does walking in an exoskeleton mean a person has recovered the ability to walk?
It demonstrates walking with that device and the support used. Recovery of the person’s own capacity is assessed separately, under appropriate conditions. Assisted walking can still be valuable when it serves a chosen goal.
Are all exoskeletons powered?
No. Passive devices use structures, springs or elastic mechanisms to support or redistribute loads. Powered systems supply mechanical action through actuators. The complete task determines whether either approach is useful.
Is a soft exosuit automatically more comfortable?
Soft construction can reduce rigid structure, but the device still transmits forces through contact with the body. Comfort depends on fit, pressure, movement, heat, controls and the activity. It needs to be assessed during use.
Do rehabilitation robots always outperform ordinary therapy?
No. Findings depend on the condition, programme, comparison and outcome. The evidence discussed here differs for walking and arm function. Ask what the proposed system has shown for people with a similar goal.
Can an exoskeleton replace a wheelchair?
It may serve some mobility or exercise goals for a suitable person. A wheelchair may remain the more practical tool for other activities. Consider the complete journey, effort, environment and help required.
Will a workplace exoskeleton prevent a back or shoulder injury?
Short-term reductions in effort or muscle activity do not prove injury prevention. Assess the whole job, any new difficulties and longer-term outcomes. Changes that reduce hazards in the work system remain important.
Does a lower EMG signal mean a task is safer?
It describes an electrical measurement under particular conditions. It does not directly measure internal tissue load or future injury. Interpret it with the task, measurement method and other relevant outcomes.
Can a rehabilitation device be used at home?
Only when the particular device, intended use, assessment, training and support arrangement allow it. Home use needs a workable environment, help when required and a plan for technical or health problems.
Will robotic assistance make muscles weaker?
The effect depends on how the device changes activity and loading over time. Assistance can enable more participation while reducing demand during a particular movement. Track relevant strength and function when these are part of the goal.
What should I ask before committing to a device?
Ask which meaningful activity it should improve, what evidence supports that use, what help and training it requires, and what the complete cost includes. A suitable trial should demonstrate the whole arrangement, including setup and recovery from problems.
Sources and further reading
Sources checked 6 September 2026. This educational guide combines selected original studies, research reviews, clinical guidance and official technical or policy sources. These sources answer different questions: a feasibility study explores a possibility, a clinical trial evaluates an intervention, and regulatory or occupational guidance addresses requirements and implementation.
The studies are examples rather than a ranking of products or an exhaustive review of every device. Findings remain attached to the population, task and support conditions studied. The practical framework and everyday examples are original illustrations.
Individual clinical suitability, occupational use and funding need assessment in the relevant setting. Device instructions, services and evidence can change. The linked sources make it possible to check the basis and limits of the article’s claims.
- World Health Organization. Rehabilitation. Fact sheet; 22 April 2024.Rehabilitation supports functioning and participation and is tailored to the person’s goals, preferences and circumstances. Return to text ↑
- World Health Organization. Assistive technology. Fact sheet, 2 January 2024.Identifies barriers to access and the importance of services around assistive products. Return to text ↑
- World Health Organization. WHO standards for prosthetics and orthotics. Standards and implementation guidance. 2017.Addresses products, personnel and service provision. Hardware selection belongs within assessment, fitting, training and continuing care. Return to text ↑
- Marchal-Crespo L, Reinkensmeyer DJ. Review of control strategies for robotic movement training after neurologic injury. Journal of NeuroEngineering and Rehabilitation. 2009;6:20. doi:10.1186/1743-0003-6-20.Explains assistance and challenge strategies. A control strategy alone does not establish clinical effectiveness. Return to text ↑
- National Institute for Occupational Safety and Health. Exoskeletons and Occupational Health Equity. NIOSH Science Bulletin. 2020.Explains how fit and representation can affect access and the physical consequences of exoskeleton use. Return to text ↑
- Vigotsky AD, Halperin I, Lehman GJ, Trajano GS, Vieira TM. Interpreting Signal Amplitudes in Surface Electromyography Studies in Sport and Rehabilitation Sciences. Frontiers in Physiology. 2018;8:985. doi:10.3389/fphys.2017.00985.Explains limitations in interpreting surface EMG amplitude as force, adaptation or physiological mechanism. Return to text ↑
- Mehrholz J, Kugler J, Pohl M, Elsner B. Electromechanical-assisted training for walking after stroke. Cochrane Database of Systematic Reviews. 2025;(5):CD006185. doi:10.1002/14651858.CD006185.pub6.101 studies; searches to December 2023. Outcomes and follow-up findings differ. Return to text ↑
- Awad LN, Kudzia P, Revi DA, Ellis TD, Walsh CJ. Walking Faster and Farther With a Soft Robotic Exosuit: Implications for Post-Stroke Gait Assistance and Rehabilitation. IEEE Open Journal of Engineering in Medicine and Biology. 2020;1:108–115. doi:10.1109/OJEMB.2020.2984429.Small study of immediate assisted performance in six people with chronic stroke; not proof of lasting recovery. Return to text ↑
- Rodgers H et al. Robot assisted training for the upper limb after stroke (RATULS): a multicentre randomised controlled trial. The Lancet. 2019;394:51–62. doi:10.1016/S0140-6736(19)31055-4.770 participants; the primary functional outcome did not favour this robotic programme over the comparators. Return to text ↑
- Lim DYL, Lai HS, Yeow RCH. A bidirectional fabric-based soft robotic glove for hand function assistance in patients with chronic stroke. Journal of NeuroEngineering and Rehabilitation. 2023;20:120. doi:10.1186/s12984-023-01250-4.Early case-series evidence on assisted hand tasks and usability; does not establish durable unassisted recovery. Return to text ↑
- Model Systems Knowledge Translation Center. Robotic Exoskeletons for Overground Walking for People With Spinal Cord Injury. SCI factsheet; accessed 6 September 2026.Guidance on assessment, training, risks and practical limitations. Eligibility and use conditions remain device-specific. Return to text ↑
- Spungen AM et al. Exoskeletal-Assisted Walking in Veterans With Paralysis: A Randomized Clinical Trial. JAMA Network Open. 2024;7(9):e2431501. doi:10.1001/jamanetworkopen.2024.31501.161 randomised participants; four months of home/community access. Use and selected health outcomes must be interpreted together. Return to text ↑
- National Institute for Occupational Safety and Health. Exoskeletons: Potential for Preventing Work-related Musculoskeletal Injuries and Disorders in Construction Workplaces. NIOSH Science Bulletin. 2022.Discusses potential benefits, implementation questions and risks. Short-term workload measures alone cannot establish long-term injury prevention. Return to text ↑
- European Agency for Safety and Health at Work. The impact of using exoskeletons on occupational safety and health. Discussion paper. 27 August 2019.Discusses benefits, uncertainties and workplace prevention. It is not certification of a particular device. Return to text ↑
- National Institute for Occupational Safety and Health. Hierarchy of Controls. Official occupational-health guidance. 2024.Prioritises controls that remove or reduce hazards through the work system. Specific interventions require task-appropriate assessment. Return to text ↑
- Slade P, Kochenderfer MJ, Delp SL, Collins SH. Personalizing exoskeleton assistance while walking in the real world. Nature. 2022;610:277–282. doi:10.1038/s41586-022-05191-1.Demonstrates personalised ankle assistance and reduced walking energy cost in the tested conditions; not a frailty or falls-prevention trial. Return to text ↑
- Kotsifaki R et al. Aspetar clinical practice guideline on rehabilitation after anterior cruciate ligament reconstruction. British Journal of Sports Medicine. 2023;57:500–514. doi:10.1136/bjsports-2022-106158.Clinical rehabilitation and progression guidance. It does not establish an added benefit from robotic equipment. Return to text ↑
- World Health Organization. WHO guidelines on physical activity and sedentary behaviour. 2020 guidelines; current recommendations checked September 2026.General activity guidance, including appropriate strengthening and balance work for older adults; not an evaluation of exoskeletons. Return to text ↑
- US Food and Drug Administration. Applying Human Factors and Usability Engineering to Medical Devices. Guidance issued February 2016.Medical-device guidance on intended users, uses and environments. Used here to explain a design principle, not to classify every fitness product as a medical device. Return to text ↑
- US Food and Drug Administration; Electronic Code of Federal Regulations. Powered lower extremity exoskeleton: 21 CFR § 890.3480. Current regulatory text; checked 6 September 2026.US special controls illustrate device-specific testing, training and labelling requirements; they are not a global eligibility rule. Return to text ↑
- European Parliament and Council of the European Union. Medical Device Regulation — Regulation (EU) 2017/745. Consolidated text dated 19 July 2026.Selected provisions on intended purpose, safety, performance and conformity. Regulatory conformity is distinct from evidence of comparative clinical benefit. Return to text ↑
- European Parliament and Council of the European Union. General Data Protection Regulation — Regulation (EU) 2016/679. 27 April 2016; selected provisions checked.Articles 4–9 and related provisions inform the discussion of personal data, processing principles, lawful bases and special categories. Dataset-specific assessment remains necessary. Return to text ↑
- European Data Protection Board. Guidelines 05/2020 on consent under Regulation 2016/679. Adopted 4 May 2020; version 1.1.Official GDPR interpretation, including employment power imbalances and genuine choice. Data-processing consent and consent to an intervention are distinct. Return to text ↑
- Kim J et al. Soft robotic apparel to avert freezing of gait in Parkinson’s disease. Nature Medicine. 2024;30:177–185. doi:10.1038/s41591-023-02731-8.Proof of concept in one person, across several conditions. Broader effectiveness requires further study. Return to text ↑
- Benabid AL et al. An exoskeleton controlled by an epidural wireless brain–machine interface in a tetraplegic patient: a proof-of-concept demonstration. The Lancet Neurology. 2019;18:1112–1122. doi:10.1016/S1474-4422(19)30321-7.Proof of concept in one participant, including suspended exoskeleton walking; not independent community mobility. Return to text ↑
- National Institute of Standards and Technology. Artificial Intelligence Risk Management Framework (AI RMF 1.0). NIST AI 100-1. 2023.Voluntary cross-sector framework covering trustworthy AI and context of use. It is not evidence that a particular fitness algorithm is effective. Return to text ↑
The best support expands what a person can choose to do.
A useful robot may help someone practise a difficult movement, complete a valued activity or preserve energy for the rest of the day. Its success belongs in that person’s life, alongside their preferences, relationships, work and opportunities.
Start with a meaningful goal, choose the support carefully and assess the result honestly. That is how engineering can become something more lasting: a dependable contribution to capability, participation and freedom.