On Wednesday 25th Feb at IDEALondon there’ll be a chance to try on two state-of-the-art exoskeletons.

​· Hilti Exo-S shoulder support suit, for reducing strain when doing overhead work

​· Hypershell X hiking exoskeleton – actively assists with hiking, running, and mountain rescue

We have no connection with either of these companies - this event is purely to get your ideas on what functions these suits do well, what can be improved, and what kind of projects we can spark as an Innovator Circle.

​There are 12 spaces available for this event but there will be other opportunities in the coming months.

Please register here if you’re interested.

Sign up here

This post will delve into:

• A biomechanical intro into gait and crouch gait - a common walking pattern for children with cerebral palsy.

• Analysis and results from patients using the Bionic Power Agilik exoskeleton

• Functional Electrical Stimulation as an alternative/complimentary approach

The information here comes from studies, simulations, and videos, including this excellent Q&A from The London Orthotic Consultancy on the Agilik exo. The Exoskeleton Report also has a great catalogue of medical and other types of exoskeletons which I’d recommend.

Gait and crouch gait

The Bionic Power Agilik suit is an active exoskeleton which applies torque at the user’s knee joint, and aims to correct or mitigate the effects of Crouch Gait (CG), this is a walking pattern found in up to 76% of children with Cerebral Palsy. Walking this way is unstable, places a lot of stress on joints, and is exhausting.

The central feature of crouch gait is walking with the knees bent, or in biomechanical terms - a fixed flexion in the knee over the gait cycle.

This is shown quantitatively in the plot of knee flexion angles in the right leg comparing standard gait (red) vs crouch gait (blue). More negative angles indicate a greater degree of bending or flexion. The data comes from experimental motion-captures of walking patterns, which are then transferred into an OpenSim musculoskeletal model.

Try walking for a few steps in this crouched gait – it’s similar to walking while doing a deep squat. This is problematic because:

• The ground reaction force acts behind your knee (effectively at the Centre of Pressure); this force along with your weight cause a moment which act to bend your knees further. To balance this your quadriceps (knee extensors) will be in constant tension.

• If we think of the knee as a hinge joint: the force from knee extensor muscles generally act ~46mm from the axis of rotation, when the knee is flexed this value – the moment arm - is smaller than the distance between the knee flexion axis from the body’s Centre of Mass and the Centre of Pressure, this mechanical disadvantage means the quadriceps need to apply a large magnitude force - leading to exhaustion and high joint contact stress.

• Walking with a flexed knee also means that the hamstrings are always in contraction – over a longer term this stiffness is ‘locked in’ (muscle contractures), and makes it harder to extend the knee.

Control strategy for the Agilik

Assuming the user has sufficient Range of Motion (RoM) in their joints, it’s tempting to think that an exo like the Agilik should apply a torque profile at their knee for the knee flexion angle to be that in a ‘standard’ gait. But how does the exo know

1. Which part of the gait cycle1 the user is currently in?

2. How much corrective torque it should apply?

Applying the incorrect torque, in the wrong direction or at the wrong time can easily lead to injuries, especially if the user is already walking with an unstable gait.

First off the Agilik has pressure sensors in each foot, this detects when one leg is supporting the user (stance phase) or whether it’s swinging through to the next step (swing phase). According to Q&A: for someone to use the suit, they need to be able to get their leg off the ground when they’re walking.

These pressure sensors also detect which part of the foot is in contact with the ground, and can narrow down whether the user has just landed their foot, through to when they’re ‘toeing’ off (entering the swing phase)

From videos of patients using the Agilik it looks like the suit:

• Applies torque to support the wearer’s knee in stance phase

• Allows their knee to flex after toe-off to swing through

• Applies torque to straighten their leg at the latter part of the swing phase

In short: my impression is that the suit is used to assist and adjust motion rather than to initiate it, the user needs to be able to walk some steps without the exo (again see the Q&A video). This also simplifies the starting and stopping of the suit.

My thoughts are that straightening the knee during the swing is safer because it doesn’t shift the user’s CoM as much, helps them to land their foot in a stable way, and requires less torque as that leg is not carrying the user’s weight. The amount of torque is possibly adjusted using feedback from the motor encoders which measure the knee flexion angle, there are also likely to be hard bounds for how much torque can be safely applied for each user, based on clinical assessments.

Results

London Orthotic Consultancy has reported great results from trialling the Agilik with patients, the key is that the suit allows users to carry on walking more efficiently outside the clinic; for the patient the suit functions like a physiotherapist which constantly corrects their gait pattern. This is especially impressive because of the variation in gait patterns with walking speed, terrain and incline. Each of these will vary the desired knee angle, and the amount of time spent in stance/swing phase. For example – when going upstairs your knees are flexed for more of the gait cycle.

From videos of patients using the suit it appears that the assist modes can transfer smoothly from walking on flat ground to going upstairs. A rehab exoskeleton like the Agilik has smaller safety margin than consumer hiking aids – where it might take ~10 steps before the assist type changes, working without manually changing assist mode is an indication of a fast feedback loop and quick adaption.

Over a longer term, it’ll be great to see data which shows that walking with legs extended can halt the ‘vicious cycle’ where hamstrings weaken with crouch gait.

FES

Again, the Agilik requires the user to be able to walk some steps without the exoskeleton; this makes the suit harder to use for patients with severe muscle spasticity2 or paralysis

Rewalk and Ekso Bionics make suits for paraplegic users, in the latter the user leans forwards on a walker to initiate gait. No doubt these companies, along with Agilik will continue to expand the number of patients who can use their exos.

Here we’ll briefly discuss another technique - Functional Electrical Stimulation – which can also be used for patients with these conditions and which I saw in action at the Exo-Berlin conference in 2025. The concept is to use electrical signals3 to recruit motor units, thus triggering muscle tension.

The advantage of FES is that:

• The user can select motion patterns without having to initiate the movement.

• The mechanical power for movement comes from the patient, preventing muscular atrophy

FES has been around for decades and has been used successfully for grasping, rowing and cycling actions. Pulse Racing is a group pushing the boundaries of this technology, this Dutch student team worked with an athlete with spinal cord injury to race in the Wings for Life competition, where he was able to cycle over 20km. See Fig 4. for an image of this setup at Exo Berlin 2025, note how similar it is to a standard gym bike.

Patients typically require months of training to before they can respond well to FES. Compared to control from an intact nervous system, artificial stimulation causes muscles to tire more quickly because the frequency of stimulation with FES is ~3x faster, and there is a different sequence to how muscle fibres are normally recruited. The patients who use FES are also likely to have some existing muscle atrophy.

With cycling the user is sat down and stable, and the movement is similar across different cycles; this is not the case with natural walking - people walk in different directions, and gait differs over uneven surfaces. Walking and balancing while upright calls for closed loop FES – to change the stimulus in real time to account for disturbances (including muscle spasticity); this requires good models of how muscles respond to activation signals, an inherently non-linear process.

A promising avenue is to combine FES with exoskeleton support, the exo can help stabilise the patient, and make up for muscular weakness in the early stages of training.

More to come on this in future posts.

1

What could possibly entice someone to wear a mechanism on their bodies, incurring extra weight, time, and possibly even ridicule?

The answer is that exoskeletons are already proving their worth by reducing injury rates, helping with rehab, and lowering the barriers for people to get active. Each iteration also brings down the weight and bulk of exos, increases force/power density, allows for better fit for different body shapes.

This series will go through three broad categories of exoskeletons:

• Workplace

• Clinical Rehabilitation

• Recreation

These distinctions are more to do with the target market than a hard classification; there’s nothing stopping people from putting on a workplace back-support exo to use around home, or for mountain rescue teams to use recreational hiking aids. One exciting trend will be the use of widespread consumer exos for rehab at home.

The focus of this post will be on workplace exos – the most commercially mature category, these are designed to protect users as they perform strenuous, repetitive tasks in their jobs. We will go through the basic biomechanical concepts behind their design, examples of commercial and academic exos, and some possibilities for future iterations.

I have no commercial connection with any of the companies mentioned in this post - the examples are chosen because I’ve come across them, or have tried their products.

Back support exoskeletons

A classic example of a strenuous, repetitive task is manual lifting. Injuries for the lower back are common because of the high contact forces involved: one study calculated lower back (L5-S1) joint contact forces to peak around 10x body weight for lifting a 11.3kg weight from knee to waist.

The forces are high because when someone bends down to lift a load, the load and upper body Centre of Mass is further away from the body’s pivot point - the lumbosacral joint - than where back muscles attach to the joint, so to extend¹ the spine the back muscles must apply a force many times greater than the load weight. The key point is that the moment arm for the load is greater than of the back muscles.

Back support exosuits are typically anchored at the shoulders and thighs, and include an element that can hold tension - e.g. a steel cable, or elastic – running down the wearer’s back; when the user bends down this element is put in tension, the tensile force pulls on the user’s trunk in parallel with their back muscles to help them extend their spine as they lift the load up.

These forces and distances are illustrated in Fig 2(a), which is taken from a paper by Tim Schubert and Robert Weidner from 2025. M_ext is the assistive torque from the exoskeleton. Note (a) shows a terrible lifting position because this pose would maximise the moment arms. F_pull in figure 1(b) shows the assistive tensile force from the exo.

Note that this tension also acts on the user at the suit’s anchor points. The key is that the suit redistributes load to avoid overstressing muscles and joints around the lower back.

What are the quantitative effects? One study measured the EMG signal2 from deep back muscles in participants performing lifting tasks, and showed a 15% reduction in erector spinae activation when the back support exo was engaged; another study modelled internal joint moments and forces with OpenSim3, and found that joint contact forces on the spine were reduced by 5-20%.

Crucially, back-support exoskeletons have been shown to reduce injuries in large scale trials, the next section will briefly go into suits offered by Herowear and Verve, which are also good examples of passive vs active exos.

Herowear

Herowear make the Apex range of passive back-support exoskeletons; passive means the suit doesn’t need an external power source; the elastic material in the suit is stretched when the wearer bends to lift – storing the energy needed to help them lift the load. This means the user expends more energy when they’re bending down. The upside is the reduced weight from motors/batteries, and removing the need to recharge.

A key part of any exoskeleton design is user comfort – work on the Apex considered how the back foam affects heat circulation, the stretchability of the thigh sleeves, and different fits for men and women. The thigh attachments were also designed to wrap securely to avoid sliding around and causing discomfort, remember that the tensile forces generated in the exo also act on the user at the attachments.

Verve Motion

The Verve SafeLift is an active back-support exo, it also has an elastic element (a ‘high tensile ribbon’) which acts in parallel with the user’s back muscles, but also contains a motor which tightens this ribbon.

There are IMUs in the thigh and upper back attachments of the suit which track the user’s motion; this data is fed used to control the motor so that it changes the tension based on what the user is doing, e.g. when they’re not lifting the ribbon doesn’t have any tension. The motion sensors also feed into a data portal which can monitor for excessive bending and twisting for workers.

Evidence

Back support exos have been trialled across several industries. A HW Apex trial with a major food retailer in the States resulted in a 47-67% reduction over baseline in strains and sprains – with zero back injuries, while a trial with a paint manufacturer found a 100% decrease in strains/sprains. For Verve a trial with a supermarket chain found zero back injuries for the duration – an improvement over the baseline. These numbers are all taken from this presentation given at the A&A wearables conference in 2025, see around 10:45 in the video for a summary.

Shoulder support exoskeletons

The same moment arm imbalance as in the lower back applies at your shoulder joint. Your arms act like a lever at the shoulder (glenohumeral) joint, its centre of mass and any load you’re holding is further away from the joint than the attachment point of shoulder muscles, so to apply a counteracting torque these muscles must exert a greater force than the weight of the load. Tasks which call for shoulder support include overhead wiring, painting, decorating, as well as automotive work – where workers stand underneath the chassis.

The Hilti Exo-S suit is an example of a shoulder support exo, it’s worn like a backpack and has an adjustable hip belt. The user’s arms rest on two supports which are each connected to a lever, the moment due to the weight from the arms and any external loads is counteracted by an elastic element (inside the case), which applies a moment via a pulley.

The position of the pulley is adjustable and changes the tension in the cord – the more tension there is, the more support you get when you have your arms raised, but because this is a passive system – the more work you then have to do to move them down.

The Hilti Exo-S is one of the few exosuits which can be purchased directly from an online store in the UK, and I’ve used mine to clean the area over my front door. Wearing the exo noticeably increases the amount of time I can do this before tiring. However, wearing the suit for extended periods causes some discomfort – as figure 3 shows the suit generates forces which act along your arms, causing a sensation of having your shoulders stretched.

More broadly, shoulder exos have been trialled at major automotive and aerospace manufacturers ; over the trial period zero shoulder injuries were reported.

Conclusions - what’s next?

In summary, workplace exoskeletons in 2026 are a practical technology, designed and increasingly proven to protect worker’s bodies in specific ways.

Industry standards are being developed so different products can be compared against an agreed-upon set of tests. These standards will need to both specify tasks which are representative of those performed in industry; and reliable, accessible ways of estimating the effects of wearing the exo. Motion capture and EMG measurements are two common techniques– but subtle changes in camera/ electrode placement can lead to large changes in readings

What’s next for the design of workplace exos? We’ve mentioned active vs passive exos in the Herowear vs Verve examples; hybrid exoskeletons combine these approaches by having a powered actuator adjust the stiffness of the support. A 2025 project from Stuttgart developed a semi-active shoulder exo which can dynamically adjust between different support torques. This keeps the responsiveness of active designs while reducing the power demand.

Looking further still, it’s an open question for me whether exos will stay specific to certain actions, or whether flexible actuators which can provide more torque for a smaller form factor can enable designs which support several different movements at once.

Please post a comment or email hello@hackingbiomechanics.com if you have any experience of using exos at work, know of other interesting exoskeleton examples, or any general thoughts. Our events will be getting started from mid-February, and there will be chances to try some of these exoskeletons for yourselves!

Share

For each step, around 200 muscles co-ordinate for the human body to move with balance and consistency, all without conscious effort.

To move freely, with control, and without pain is critical for a good quality for life; and we are sorely in need of improved ways to increase the lifespan over which we can do this.

This will be an Innovator Circle for exploring the science of human motion, and the technology for maintaining it in the face of the muscle loss and joint degeneration that comes with ageing.

Biomechanics is both an established and emerging field; over centuries anatomists have mapped out the muscles, tendons and ligaments of our bodies, yet we continue to be surprised by the ways they link together in movement. Researchers are actively investigating the lifting postures which lead to greater lower back stress, and the causes of disproportionate ACL injuries in elite female footballers.

The toolkit for analysing biological motion has come a long way since Muybridge’s iconic photographs of a horse in gallop. A host of technologies has enabled this: cameras with higher frame rates and resolution for motion capture, piezoelectric sensors for force detection, neural networks for mapping sensor readings to joint angles – to name but a few.

More understanding leads to improvements in health outcomes. Knowing the prime drivers for gait allows clinicians to choose the right muscles to target in rehab, understanding the mechanisms of ankle sprains has led to orthotics which activate just in time while remaining unobtrusive; better gait prediction allows exoskeletons to assist our movement more naturally.

The core of this Innovator Circle will be small groups from academia, industry, open source communities and clinicians who meet regularly to scope out new ideas; our bet is that the potential in biomechanics is such that this light-touch approach can spark projects with tangible outcomes.

Most of all you would be someone who’s keen to share ideas that will bring about more accurate and seamless mocap for real clinical settings. Please register your interest here if this could be you.

We will also be sending out regular posts on this field, plus holding talks, demos and hackathons in London from February 2026. If you’re generally curious about biomechanics, exoskeletons, wearables or rehab tech, please subscribe for updates