This robot looks a bit like a dog and can be badminton

A game of badminton requires quickly footwork, good hand-eye coordination and sharply estimating the shuttle movement. For robots this is a challenging combination of tasks, but researchers from Technical University Eth Zurich in Switzerland have developed a robot that can take a nice shuttlet. She demonstrate the robot skills in Science Robotics.

The badminton robot looks a bit like a dog: he has four legs that come together in a wide ‘body’. A robot arm is mounted on the spot where you would expect a head. Vast, the badminton racket that is attached to the end comes above the net. The robot only uses equipment in its own body, there are no external cameras or computers that give it signals.

The challenge is to let the legs work well with the arm, which at the right time has to make a fast and powerful movement to get the shuttle just over. The four legs can move separately and each have a hip joint and knee joint. The arm has three hinges that can go up and down and turn around their axis.

Good visibility is important to be able to estimate what the shuttle will do. People have a wide field of vision, can quickly focus and image processing is super fast in our brains. Robot technique has improved in this area in recent years, but there is still a lot to be gained here, the researchers think.

Predictive model

Multiple -knotted algorithms control the movements, a predictive model for the shuttle movement and self -learning models for the control of the legs and the arm. These algorithms must always make considerations with each other: does the whole body move towards the shuttle, or is the arm extended further? A vast arm also means something for the legs, because the robot has to stay in balance. To promote speed and cooperating movement of legs and arm, the researchers have more integrated the algorithms than was previously usual.

The robot managed to have a rally of 10 strokes play with a person, outside, with mild wind. Video images made in the lab do suggest that a good amateur is easily beating the robot. The robot is not yet playing with the backhand, the researchers hope to teach him in the future.

A sporting robot is not the final goal but a challenge that roboticians like to set themselves. They can learn their robots step -by -step skills that also come in handy if they ever have to function in the complex real world. Sport is a defined yet challenging version of that real world.

Table tennis has already been extensively investigated, so preliminary work had already been done in terms of technology and algorithms. Badminton is more challenging, the researchers write, because the playing field is larger and requires the more movement and cooperation of the entire body.

To share

Mail the editors