LIAM GALLAGHER, formerly of the band Oasis, tends to stroll with a roll to his shoulders. John Wayne’s slow swagger has been linked to everything from a misaligned leg to small feet. Some say Vladimir Putin’s distinctive shuffle is thanks to KGB weapons training that encouraged operatives to dampen the swing of one arm to keep it closer to their gun.
Considering that walking is such an everyday function of a bipedal species, it is incredible that we find so many different ways to do it. Perhaps that’s why our gaits – and what they say about us – are so fascinating. It takes dozens of muscles working together throughout the body to put one foot in front of the other. These subtle patterns of muscular flexes and strains are highly distinctive, so much so that scientists who study gait increasingly believe they are as unique to you as your fingerprint.
Advertisement
Gait analysis has been around for years, but now it is going mainstream. China is using it to track its citizens. Transport companies want to use it to identify ticket holders. Doctors say an analysis of your strides might provide an early hint of health problems. But is this technology on a solid footing? And is it offering a step in the right direction or is it merely another worrisome invasion of our biometric privacy?
We have watched other people walk for centuries. The ancient Greek philosopher Aristotle was one of the first to pay attention, but no one was more obsessed with the subject than the 19th-century French novelist Honoré de Balzac. He peppered his books with extravagant descriptions of how his characters got from A to B, often linking their walking style to their personalities, social status and occupation. Describing an elderly woman in his 1832 book The Vicar of Tours, he wrote that “her movements were not equally distributed over her whole person, as they are in other women… She moved, so to speak, in a single block.”
The Paris strut
Balzac also produced one of the first serious examinations of gait. “It seems, the science is mine,” he wrote on realising the topic had been largely ignored. His Theory of Walking, published almost 200 years ago, was based on watching people as they strutted the streets of Paris. He made some pertinent observations: sailors, who were used to counteracting the rocking motion of the sea, would move with their legs wider apart. It wasn’t science as we know it, but it showed beyond doubt that walks vary a great deal.
Balzac doesn’t have the field to himself any longer. Gait analysis has grown as an academic pursuit and is producing practical applications. Among the first to spot the potential were criminal investigators, who have used to identify a suspect from the way they walk in CCTV footage.
More recently, advances in computer vision have allowed gait analysis to make big strides, using precision imaging to measure details the eye can’t perceive. Some gait scientists record and compare specific biomechanical measurements, such as how far an ankle rotates, or how large the angle between the upper and lower leg gets as the knee flexes. Computerised gait analysis looks at the total set of such differences. “Everyone has a unique gait pattern. And that, of course, means they could be identified,” says , who studies gait at Kiel University in Germany.
“One of the biggest breakthroughs in the early days of gait recognition was realising that you can recognise someone by their silhouette,” says Patrick Connor, a gait-recognition researcher with the company Stepscan Technologies in Canada. He is talking about what is known as gait energy imaging. This involves taking a series of snapshots of a person’s profile as they walk, usually from the side, and merging them into a single image, each unique to their owner. “The head’s in the same place and the torso’s in the same place and you get this kind of blurry arms and legs movement,” says Connor.
How reliable is it? Gait researchers say a system “works” if it can correctly identify 90 per cent of volunteers from their walks. As the technology improves, they test it against bigger and bigger groups. “When I started, we could just about identify 10 people,” says , UK. “These days, people are getting 90 per cent on enormous databases of tens of thousands of people.”
It is also beginning to work with less than ideal imagery. In 2017, computer scientists in Portugal showed that they could . This may mean drones can carry out covert gait analysis – on sunny days, at least.
Secret stalking
All this might prompt thoughts of face-recognition technology, which is becoming widely used to identify people without their knowledge in many countries. In the UK, police use of this technology has been subject to a court case in which a man claimed that his human rights had been breached. Gait analysis could be an even more powerful tool because it can work with lower resolution pictures and is much harder to avoid than face recognition systems. You can generally fox these by wearing a mask – which wouldn’t exactly look suspicious these days.
Paul Wiles, the , points out that people upload lots of video clips to social media where faces aren’t easily made out, but gaits are. He says police in the UK are interested in pairing gait analysis with face recognition to make better use of this kind of footage, though this is on the back burner at the moment due to the coronavirus pandemic.
Gait analysis is nowhere near as widely used as face recognition yet. This is mainly because it requires a series of sequential pictures that cover the seven stages of walking, as each leg moves and lifts. Processing these images requires serious computing heft and most gait analysis systems don’t have enough of this to work in real time. This is why these systems are of limited use to police; a crucial requirement for use in law enforcement is that the systems can recognise those they want to apprehend quickly enough to give authorities a shot at detaining them.
The one place where gait analysis is being used a lot is China. In late 2018, that state authorities had extended the biometric surveillance of its population to include this type of analysis. Huang Yongzhen, the CEO of Watrix, the tech firm that developed the Chinese system, said at the time that it can identify people up to 50 metres away with 94 per cent accuracy. Wanted criminals should watch out, Yongzhen warned: the company’s database includes many of their distinctive walks.
Other experts in the field are sceptical that Watrix’s technology is as powerful as it claims. (The company didn’t respond to questions from New Scientist.) But most agree it is only a matter of time before such real-world identification is possible.
For Wiles, that is worrying. He has said that the use of gait analysis by police and others might . Such biometric analysis is ethically problematic because it can be done without an individual’s knowledge or consent, he said, and because “some uses of new technology may… not be in the public interest nor justified by the benefits”.
“Gait analysis could be an even more powerful tool than face recognition”
Wiles says his preferred way forward would be for governments to draw up frameworks that set out the broad principles for using gait and other biometric identifiers. The Scottish government has already started down this road by introducing .
All this applies to situations in which people don’t necessarily want to be identified. But there are upsides. For those looking to be recognised, gait analysis offers advantages. Take airport security, where rapid, accurate identification is a must. Gait recognition could simplify the whole process.
If people want to be identified, there is no need to go to the trouble of measuring silhouettes, as in gait energy imaging. Instead someone might be asked to wear sensors that would feed back precise information about their movements. These needn’t be anything too sophisticated. Computer scientist at the Sapienza University of Rome uses this method to recognise people from the information sent from the as it jerks around in their hand or pocket as they walk. Her team has built a prototype smart door that opens when it recognises an approaching person in this way.
In principle, this is much more convenient than, say, pausing to scan an iris or fingerprint at a checkpoint. “Naturally, that’s going to slow down the amount of traffic that can flow through any particular entry point,” says Connor. “But if you have to walk up anyway to the access point, you’ve already presented your gait. There’s nothing more to do.”
Several firms appear to agree. Mastercard says it is working with transport companies on how . Nuclear energy officials in Indonesia want to .
Others are taking an even simpler approach by comparing not patterns of muscle movement, but footsteps. “When we see a footprint in the sand, we see an image. But there’s actually a sequence of events that occurs during the footstep and a certain amount of pressure that’s applied in certain locations,” says Connor. The way these patterns change from person to person is already used to help design and fit sports shoes. By fitting pressure sensors inside floor tiles, scientists can read and compare these signatures, so our footsteps can open doors.
One concern here is “spoofing”, where someone analyses and impersonates a walk in order to, say, get inside a secure location. “I think that probably the biggest concern is whether someone can steal your identity and use it for their own purposes,” says Connor. It remains to be seen how vulnerable gait-recognition systems are to this sort of attack.
“Could someone impersonate your walk to get inside a secure location?”
Rate my gait
Once scientists, companies and the police have your gait information, what could they tell about you, beyond your identity? Just as Balzac linked walking styles with temperament two centuries ago, people often claim to be able to judge a person by their gait. Whether this can be done is controversial, however (see “What your walk says about you”).
More usefully, there is growing interest in using gait analysis to track people’s health. Heinzel uses sensors fitted to the lower backs of people with Parkinson’s disease to analyse their walking movements. Videos of people walking are already routinely used to spot muscular problems, but Heinzel hopes to use this technique to diagnose the disease, track physical deterioration and assess how much drugs are helping.
It is early days, but he says the approach could be used to signal the onset of other conditions too. “How you move in a supermarket may change when you develop mild cognitive impairment or when you have diseases like multiple sclerosis,” he says. Catching such conditions early would allow better treatment and could mean a better prognosis.
Knowing precisely how you walk can also help you do it better. Kyle Reed, a biomechanics researcher at the University of South Florida, uses the same body-sensing technology and cameras used to make the film Avatar to examine the impaired gait of people who have had a stroke. This can improve rehabilitation programmes, he says, because the technology adds information that a physiotherapist can’t see. “There’s a gap right now between what therapists do sort of intuitively and what the computer models are saying,” he says. “One of the things I’m trying to do is bridge that gap.”
His research has confirmed what Balzac knew all those years ago: there is more than one way to walk. The writer couldn’t have foreseen that what he observed from the Paris pavements might one day provide a means of identifying every individual on a planet of nearly 8 billion people. Had he lived to hear of it, perhaps he might have walked with a little more swagger.
What your walk says about you
It is a common belief that the way someone walks offers clues to their personality. A long, quick stride is often believed to betray confidence or arrogance, while a shuffle is a sign of an introvert. But are gait and personality really linked?
In 2012, researchers at Durham University in the UK asked volunteers to judge the personality of 26 people as they walked. Although the volunteers agreed with each other – that someone who walked with a loose, expansive style was an adventurous, trustworthy, warm extrovert, for example – they .
Still, some studies do find a link. In 2017, psychologists at the University of Portsmouth, UK, looked for correlations between someone’s personality and biomechanical analysis of their gait, such as how much they swayed. They found some: . As their left foot went forward so did the right shoulder, in what has been called a “pimp roll” or the “gangster glide”.
Ask the inmates
In a separate study, Canadian psychologists who had been imprisoned for violent crimes. Each student had been asked by the researchers to say how often they had felt “victimised”, with the definition left intentionally broad.
The prisoners who scored highest on a test of psychopathic traits were better at telling from the students’ walks which had said they had been victimised most often.
That leaves us rather uncertain as to what your walk really says about you. Most psychologists reckon there is no link between personality and walking style. But studies on this topic are so small and varied in their design that it is tough to confidently draw any conclusion.
