NSK has developed a four-legged robot with wheels, and are working towards the creation of a robotic guide dog.

“This robot has wheels on the ends of its legs, so basically, on flat terrain, it uses the wheels to move. In places like stairways where the robot can’t move on wheels, it can use its legs instead.”

“Raising a guide dog takes a long time, and a dog can only work for about 10 years. So the number of guide dogs can’t be increased instantly. Also, some people don’t like having animals around. So we’ve developed this robot, as a way of doing similar things to a guide dog.”

The robot has a handle with a variable angle and length, so the user can maintain a stable posture and consistent grip at all times, even when being guided downstairs.

“The grip has a force sensor at the end, to enable the robot to move when instructed. This is used in an intuitive way, so if you grip it facing forwards, the robot goes straight, and if you grip it so as to twist it, the robot turns.”

To recognize steps, the robot processes images from a sensor. The sensor data is converted into shape, position, and attitude information in 3D space. This enables the robot to recognize the number of steps and their width. It makes recognition possible even while going up stairs, which was difficult before.

“There are proximity sensors on the ends of the legs. If we put a sensor on the head, the robot can see what’s ahead, but it has blind spots around the legs. So the distance image sensor on the head is used to recognize steps overall, while information around the legs is obtained using the proximity sensors on the legs.”

From now on, NSK plans to do field tests on safety, with a view to releasing a commercial version.

“We haven’t considered the safety aspects very much yet. We’ve finally reached the level where the robot can recognize and climb stairs. So what we want to think about now is safety; for example, how to avoid falling, how to recover and keep climbing if a fall does happen, and how to prevent the user’s fingers from getting caught when the robot moves.”

This Video is provided to you by DigInfo.tv, AkihabaraNews Official Partner.