Robotic gait training

A motorised exoskeleton is fitted to the legs above a treadmill, and a harness carries a set proportion of body weight. The robot then drives the hips and knees through a walking pattern while the treadmill moves underneath. Body-weight support, walking speed and how much the robot assists are each set separately; as the patient improves, support and assistance are reduced and speed is increased.

The point is repetition. Walking is relearned through many correct steps, and a therapist manually assisting a patient who cannot bear weight becomes physically exhausted well before enough repetitions are done in a session. The robot keeps delivering steps without the pattern degrading as the session goes on.

Patients who cannot yet walk safely or at a useful volume unaided: after stroke, spinal cord injury, traumatic brain injury, in multiple sclerosis, and in Parkinson's disease where gait has deteriorated. It is most valuable in the window where a patient has some potential to walk but not enough strength or control to practise conventionally. It does not suit someone already walking reasonably well — for them, floor-based task practice transfers better.

It requires sufficient joint range at hip and knee, no unhealed fracture in the legs, skin intact where the cuffs sit, and cardiovascular stability for upright exercise. Bone density matters in long-standing spinal cord injury, where fracture risk during loading is real. These are checked at assessment, and a patient who does not meet them is told so rather than fitted anyway.

Robotic gait training combined with conventional physiotherapy improves the odds of independent walking after stroke. In the 2025 Cochrane review of 101 randomised trials with 4,224 participants, adding it to physiotherapy raised the odds of walking independently (odds ratio 1.65, 95% confidence interval 1.21 to 2.25; 51 trials, 2,148 participants; moderate-certainty evidence) — about one more person walking independently for every nine treated 1. People who were not yet walking, and those in the first three months after the stroke, appeared to benefit most 2.

The same review found no meaningful gain in walking speed or walking distance, and whether the benefit lasts after the programme ends is uncertain 1. It is not clear that it beats an equivalent amount of skilled conventional gait training in patients who can already walk 2. We do not present the robot as superior to a physiotherapist, and we do not build a programme around device time alone. It is a tool that lets us give more of the right practice, not a replacement for the practice.

The Lokomat is a treadmill-based robotic exoskeleton that supports body weight and guides the legs through a walking pattern. It allows gait training to begin before independent standing is possible, and it records every session so progression can be measured.

The patient is supported in a harness over a treadmill with the legs attached to powered orthoses. Body weight support and guidance force are set by the therapist and reduced as ability improves. A session typically runs 30 to 45 minutes.

Body weight support commonly starts high and is reduced as tolerance improves, as is the robot's guidance force. Those two numbers, recorded each session, are a more honest record of progress than the number of sessions completed.

It produces a symmetrical, guided pattern, which is useful early and limiting later. Real walking involves variability, obstacles and reacting to being off balance, so over-ground and outdoor practice has to follow.

There is also a risk of the patient becoming passive inside the device, letting the robot do the work. Modern systems vary the assistance to counter this, and a supervising therapist watching for it matters more than the setting.