Play badminton – with a robot – Measure + test

A research team from the ETH Zurich taught the four -legged robot anymal to play badminton game – including a precise arm swing, active perception and skillful footwork.

▶
Listen to this article

How do you get a robot to run, see and strike a badminte back at the same time? Researchers at ETH Zurich under the direction of Marco Hutter, professor of robot systems, have investigated this question. They developed a control that coordinated leg movements, arm guidelines and camera views.

The robot anymal persecutes the shuttlack with two cameras, estimates the trajectory and runs to the right place to strike at the right moment.

In the video, Andrei Cramariuc, research assistant at Hutter, tells how the training went and what challenges there were to master.

Job offers+
Suitable for the topic

© ETH Zürich / Youtube

Researchers at ETH Zurich taught a robot to play badminton games.

---

Read more about the topic

---

More articles on ETH Zurich

More articles on robots

More articles on robotics

Key Data Points: Anymal’s badminton Performance

To further illuminate Anymal’s remarkable badminton skills, consider this comparative table:

| Feature | Anymal (ETH Zurich) | Human (Elite Badminton Player) | Key Metric |

| ——————– | —————————————————— | —————————————————————- | ——————————————————————————————————————————————————- |

| Movement Speed | Up to 2 m/s (dependent on court size and shuttlecock location) | Up to 7 m/s (shuttlecock can reach speeds up to 300 km/h) | Agility and reaction time are crucial for both. Anymal’s speed is improving with ongoing software updates. |

| Vision System | Integrated stereo cameras with trajectory prediction | human eyes and brain, capable of complex visual processing. | Accuracy of shuttlecock tracking.Anymal utilizes sophisticated algorithms in real-time. The accuracy depends on lighting conditions. |

| Swing Accuracy | Moderate, influenced by environmental factors | High, fine-tuned through years of practice | Consistency of hit and placement. Further refinement of Anymal’s arm control system to optimize each shot is anticipated. |

| Footwork | Dynamic leg movements, adjusted in real-time. | Precise, speedy footwork for optimal court coverage. | Court coverage efficiency and responsiveness to shuttlecock direction. Anymal integrates complex algorithms to estimate the shuttlecock’s path. |

| Processing Time | Real-time, Milliseconds for Calculation and Adjustment | Sub-second for Decision-making, with milliseconds reaction time. | Speed of details processing and response time to shuttlecock. This includes data acquisition, calculation, and physical robot movement. |

| Adaptability | Limited, requires specialized programming | Extraordinary, can adapt to changing game scenarios. | ability to learn and adjust strategy.This relies on data-driven learning and algorithms continuously improving Anymal’s strategies and gameplay.|

Note: All values are for demonstration purposes and subject to ongoing research. The comparison considers factors like reaction time, and the precision with which they hit the shuttlecock.

Frequently Asked Questions (FAQ)

Q: How does Anymal “see” the shuttlecock?

A: Anymal uses two high-resolution cameras – a stereoscopic vision system, which is capable of calculating the trajectory of the shuttlecock. The system processes visual data to estimate the shuttlecock’s position, speed, and predicted path.

Q: What kind of control system does Anymal use for playing badminton?

A: Anymal utilizes a sophisticated control system that coordinates its leg movements (for running and positioning), its arm swing, and the direction of the cameras providing it with an accurate and calculated response.

Q: What are the biggest challenges in teaching a robot to play badminton?

A: Major hurdles include real-time trajectory prediction, precise coordination of movement and arm swing, and the ability to adapt to an unpredictable surroundings.This also includes the time required to process information.

Q: how does Anymal’s performance currently compare to a human badminton player?

A: While Anymal can successfully hit the shuttlecock and react to its path, its performance is still noticeably less then a professional badminton player. While it may be more accurate in its calculations, its physical movements, speed, and adaptability still need substantial improvements.

Q: What are the future implications of this research?

A: This research contributes to advances in robotics, including improved robot perception, navigation, and complex motor skills. Applications extend to search and rescue, automated manufacturing, and the creation of sophisticated, adaptable general-purpose robots.

Q: Is Anymal learning to play Badminton on its own?

A: The robot is trained and programmed at ETH Zurich. Deep learning is a key component of Anymal’s growth,along with its perception and complex algorithms. By playing against itself and gathering data, the robot is continuously improving its play. The development team at ETH Zurich meticulously programs Anymal with the knowledge required to play.