facebook link iconInstagram link iconyoutube channel link iconlinkedin link icon
email and contact iconicon for location

Google Robotics Learn Ping Pong

The Robotics team at Google has taught a robotic arm to play 300+ ping pong shot rallies with other people. While you may not be a big table tennis fan, the same techniques could be used to train robots to perform other “dynamic, high acceleration tasks” that require close human-robot interaction (like a surgical assist maybe).

Table tennis was chosen because it requires both fast and precise movements in a structured game that occurs in a fixed and predictable environment - this makes it a great set up for exploring human-robot interaction and reinforcement learning techniques (where the robot learns from doing).

What Does It Solve?

What problems does this model of robot learning really solve? The team explained that if you don’t have a good robot policy (a set of rules for the robot) to begin with, then you can’t collect high-quality data on how people will interact with it. But, without a human behavior model to start with, you can’t come up with the robot policy in the first place. It's the old "chicken-or-the-egg" dilemma. Their alternative solution was to train a robot in the real-world from human interaction – a process that could take a long time (and people can get hurt by swinging robotic arms).

Researchers sidestepped this limitation by using a simplified model of human behavior as a starting point and then trained the robot both with a simulation and a by interacting with real human. With each subsequent iteration in the process, both the human behavior model and the robot policy were refined.

Teaching robots to play games might seem trivial, but addressing these kinds of robot training problems potentially has real-world applications. What other tasks could robots learn faster and better from interacting dynamically with humans?

Even if Googles current ping pong robotic project only leads to a future table tennis robot “coach” able to adapt its play style according to the skill level of the human student – it’s a testament to how fast the field of robotics and machine learning are progressing.

See Full Article Table Tennis: A Research Platform for Agile Robotics

Robotics In Orthopaedic Surgery

Thank You for Subscribing to Bone Health & Harmony Blog!
Oops! Something Went Wrong, Please Enter Your Email Again.
Cory Calendine, MD is an Orthopaedic Surgeon and founding partner of the Bone and Joint Institute of Tennessee at Williamson County Hospital in Franklin, TN. Dr. Calendine is an expert in Joint Replacement, specializing in Hip and Knee Surgery. From diagnosis through treatment, the Orthopaedic Surgical experts at the Bone and Joint Institute use the latest techniques and technology to improve care for people with musculoskeletal problems. For more information, please contact our office or schedule your appointment today.  

Vitamin D, Bones And COVID-19

Vitamin D is important for good muscle health. People with very low Vitamin D blood levels are more likely to experience muscle cramps, bone or joint pain, and may be at higher risk of complications from COVID-19. Vitamin D is vital for healthy bones, muscles and immune system. Studies have confirmed that older people who get adequate Vitamin D have fewer falls, fewer bone fractures and better muscle function.

Read More

Artificial Intelligence and Knee Osteoarthritis Treatment Decision Making

Artificial Intelligence may offer decision-making assistance for patients dealing with joint osteoarthritis. New medical research has demonstrated that artificial intelligence augmented treatment decision tools offer benefits over more traditional digital patient education materials.

Read More

Common Causes of Knee Pain in Women

Knee pain affects many adults, especially after the age of 50. Dr. Cory Calendine, MD reviews the more common causes of knee pain seen in women including early onset osteoarthritis, degenerative meniscus injury and patellofemoral pain syndrome.

Read More