Training guide dogs is a laborious and time-consuming process. Not every dog is able to do this, so the animals have to be carefully selected and trained individually. It's easier with a robot: if you develop one general algorithm, as many electronic dogs as they like can work on it. Four-legged robots have improved a lot over the past few years and at the same time have become more affordable. And since they all look like dogs, the idea of using them as guides seems logical.
UC Berkeley has created a guide dog based on a small four-legged robot called the Mini Cheetah. It is equipped with a 2D lidar - a sensor for sensing the environment, as well as a gimbal camera for tracking a person's position, a leash for guiding people and a force sensor on a leash - it measures the force applied by a person to a person. The invention is reported by TechXplore.
A person keeps such a dog on a flexible leash that can be tightened and loosened - this provides flexibility in the interactions between man and the "animal". Scientists have developed signals that can make communication easier. For example, when a person and a dog find themselves in a narrow space like the corner of a corridor, the robot can first loosen the leash to change position and find the correct orientation, and then pull on it - this will be a signal for the person to resume movement.
Scientists have tested their robotic guide dog in a series of experiments. The robot had to guide the blindfolded person to a certain place, avoiding collisions with obstacles. The test results were very promising. Now engineers are training their robotic guide to navigate the open air and guide people through the cross with traffic lights.