Serving technology enthusiasts for over 25 Years. TechSpot is a trusted source for tech advice and analysis.
What it means: Since Boston Dynamics’ famous legged robots were first displayed in 2013, a number of organizations have developed machines that perform professional and recreational tasks autonomously. The latest research uses badminton as a way to improve a robot’s coordination and agility, which could lead to improved training methods and maneuverability.
Scientists from ETH Zurich published a study recently. Study as well as video (below), explaining how they taught a quadrupedal robotic to play badminton. The research could show the value of a full body training model which does not break activities down into individual steps. This could potentially increase the range of tasks that can be learned.
An ANYmal-D standard robot was fitted with a DynaArm, and taught using an reinforcement learning model that required minimal software or hardware expertise. The machine also learned to track the shuttlecock by using its stereo camera.
The robotic learned to navigate the court by using Nvidia’s Isaac Gym Virtual Simulator, which took approximately 7,500 iterations in almost five hours on an RTX2080 Ti graphics card. A noise prediction model also helped it track shuttlecocks that passed outside the camera’s field of view.
It eventually learned to use various complex movements, such as starting gestures, backswings and hits, running and recovery, to hit the shuttlecock, without the researchers having to teach them separately. It adjusted its movements according to the shuttlecock’s perceived velocity, galloping across court in order to hit it on time.
This experiment shows the adaptability and versatility of the ANYmal D, which ANYbotics developed to perform inspections, respond to emergencies, and other tasks within various industrial environments. The company’s website shows the robot operating around the clock while navigating indoor facilities. It can navigate gravel, stairs, caves and snow. The ANYmal-X is a new model that has been certified for hazardous environments. It can be used to conduct inspections in oil, gas and chemicals.
Google DeepMind has trained a robot to play table tennis. Engineered Arts has showcased a humanoid artist robot, and the NBA has recently deployed robots for athletes to practice. Chinese companies have also deployed robots in recent years to compete in kickboxing and a half-marathon with mixed results.
