Could a Robotic Guide Help Blind Travelers Navigate Public Spaces?

Chieko Asakawa, a blind computer scientist, regularly flies between the US and Japan. At both ends she often has to wait for an airport escort. The waiting is annoying. Giving up control of your own route to someone else is worse. 

Why airports are hard for blind travelers

Think about what an airport asks of you. Follow the signs. Find your gate on a screen. Dodge a luggage cart, a family of six and a floor cleaner, all in the same minute. None of that comes easily when you can't see the signs. Escorts help, and they're kind about it, but you end up running on somebody else's schedule. 

The usual tools are a white cane and a guide dog, and both work well for the people who use them. The trouble is numbers. A guide dog means a residential training program, and one guide dog breeder's figures put ownership at around 2 percent of blind and partially sighted people in the US. About 98 percent use neither a dog nor a cane, according to one news report. For most people, the question of what else could help is a real one.

CaBot and the AI suitcase

Asakawa's answer came from her own luggage. What if the suitcase could do the guiding? She began working on that idea in 2017 at Carnegie Mellon University. The result, CaBot, short for carry-on robot, looks like an ordinary suitcase. Inside are cameras and sensors similar to the ones on a self-driving car. You give it a destination through a phone app, it plans a route, and the handle vibrates to tell you which way to turn. 

One design scenario from a student team shows how it might feel. A traveler says food, and the robot reads out the restaurants nearby. She picks one, the suitcase walks her over and announces when she's there. 

Ten blind users tried an early version. It beat their expectations, and many compared it to walking with a guide dog or a sighted guide. They reported high confidence, a sense of safety and plenty of trust. The idea has since grown into the AI Suitcase, a research platform with LiDAR, depth cameras and a detailed indoor map. It's been tested in museums, shopping malls and airports, including Pittsburgh's and New Chitose in Japan. 

Part of the charm is how ordinary it looks. Nobody stares at a suitcase. The catch is maps. One early prototype leaned on an airport floor plan that had been mapped in advance, plus Bluetooth beacons dotted along the halls. A related suitcase called BBeep, built by researchers from Carnegie Mellon, the University of Tokyo and Waseda University, targets crowds.

Glidance glide

Glide, from a company called Glidance, takes another shape. It looks like a small vacuum cleaner with a handle and two wheels. You push it ahead of you, and onboard sensors, cameras and AI steer the wheels around hazards. The handle carries the feel of the ground back to your hand, and you set the walking pace. 

Glidance founder Amos Miller is blind and uses a guide dog himself, and he describes Glide as something like an electronic guide dog. The practical details matter. The handle folds so the device can ride on a bus or a plane, and learning to use it takes far less time than a guide dog program. In research, two control styles were tested: Glide leads, or the user leads. People found both easy to learn. 

Reporters who tried it at a trade show saw its wheels steer around people in a crowded hallway. When one reporter stood still in its path, it stopped instead of rolling forward.

Robot guide dogs

Some researchers prefer legs. At Binghamton University in New York, a team used reinforcement learning, a trial-and-error training method, to teach a four-legged robot to lead a blind person around obstacles in a lab hallway. It picked the skill up quickly. In theory, legs could handle stairs and rough ground better than wheels do. 

Legs also draw more looks. Miller has pointed out that a four-legged robot carries some sci-fi baggage a plain wheeled device doesn't, and being noticed is not what everyone wants on a quiet morning commute.

Robots vs. Canes and Guide Dogs

None of these machines is meant to replace what already works. A cane reports the ground right in front of your feet. A dog reads traffic and steps around trouble before you notice it. Robots add something else: a map, a destination and a route in places you've never been. For someone without a guide dog, which is most people, that could fill a gap. 

Control matters too. Researchers studying Glide found that how much control a device gives its user shapes trust and confidence. A robot that grabs the wheel completely may feel less like help and more like being steered. People may prefer a guide, human or machine, that lets them set the pace. 

Nobody has built the perfect guide yet, and maybe nobody will. But a suitcase that vibrates left at the right moment, in a gate area full of strangers, is a good start for a blind traveler.