Robot guide dogs: What do they need to be able to do?
There are many studies on robotic guide dogs. However, none have yet made it to series production. Researchers have now systematically approached the topic.
A robotic guide dog can safely navigate visually impaired people.
(Image: Stephen Folkerts / Binghamton University)
Guide dogs can help visually impaired people regain a certain degree of autonomy. However, there are not enough trained guide dogs, they are expensive and in many cases they cannot be adequately cared for by their users. This is where robot guide dogs come into play. In a scientific study, researchers at the University of Massachusetts Amherst have investigated what expectations they need to meet.
"There are 40 years of studies and none of these robots are actually being used by end users. We tried to address this problem first so that before we develop the technology, we understand how they [the visually impaired] use the guide dog and what technology they are waiting for," says Donghyun Kim, a professor in the UMass Amherst Manning College of Information and Computer Science and one of the study authors involved.
The aim of the study "Towards Robotic Companions: Understanding Handler-Guide Dog Interactions for Informed Guide Dog Robot Design", published in Proceedings of the CHI Conference on Human Factors in Computing Systems, was therefore to find out the requirements placed on robotic guide dogs.
To this end, the research team conducted qualitative interviews and observation sessions with 23 visually impaired guide dog handlers and five trainers of this type of dog. The researchers analyzed the interviews to find out what limitations guide dogs have, what expectations are placed on them, and what conclusions can be drawn for the development of robotic guide dogs.
Visually impaired people want to control robotic guide dogs
One important aspect that emerged was the question of the balance between the autonomy of the robot and human control. The scientists had originally assumed that a robotic guide dog would have to function in a similar way to an autonomous car. In other words, the robot navigates the blind person to where he cannot get to on his own. However, as the researchers learned from the interviews, this takes away too much of the visually impaired person's own self-determination. Instead, blind people want to be able to control the entire route. The dog should only be used to navigate around local obstacles. In addition, a guide dog must nevertheless be able to head for certain destinations independently.
The scientists have also identified a number of very practical criteria that a robot guide dog should fulfill. These include running time, for example. This was important to around 90 percent of respondents, as current four-legged robots only have a battery life of less than four hours. It is also important that the robot dog uses several camera angles to be able to recognize obstacles that are above the head of the visually impaired person, for example. Audio sensors could also prove helpful, for example to detect dangers approaching from hidden areas. The robotic guide dog should also be able to help people board the right form of public transport and find a seat.
The scientists see their scientific work with 2000 minutes of audio and 240 minutes of video material as an initial starting point for deriving further criteria for a robotic guide dog.
The jury of the Conference of Human Factors 2024 agreed. The paper received the "Best Paper Award" at the conference. This puts the study in the top one percent of all scientific papers submitted to the conference.
(olb)