Enabling humanoid robot movement with imitation learning and mimicking of animal behaviors

Enabling humanoid robot movement with imitation learning and mimicking of animal behaviors

Rish Joshi

Rish is an entrepreneur and investor. Previously, he was a VC at Gradient Ventures (Google’s AI fund), co-founded a fintech startup constructing an analytics platform for SEC filings and labored on deep-learning analysis as a graduate scholar in pc science at MIT.

More posts by this contributor

Over the previous 20 years, humanoid robots have drastically improved their capability to carry out capabilities like greedy objects and utilizing pc imaginative and prescient to detect issues since Honda’s launch of the ASIMO robotic in 2000. Despite these enhancements, their capability to stroll, soar and carry out different complicated legged motions as fluidly as people has continued to be a problem for roboticists.

In current years, new advances in robotic studying and design are utilizing knowledge and insights from animal habits to allow legged robots to maneuver in rather more human-like methods. 

Read More:  Microsoft makes Teams video meetings less tiring with its new Together mode

Researchers from Google and UC Berkeley revealed work earlier this yr that confirmed a robotic studying methods to stroll by mimicking a canine’s actions utilizing a way known as imitation studying. Separate work confirmed a robotic efficiently studying to stroll by itself via trial and error utilizing deep reinforcement studying algorithms.

Imitation studying particularly has been utilized in robotics for varied use circumstances, equivalent to OpenAI’s work in serving to a robotic grasp objects by imitation, however its use in robotic locomotion is new and inspiring. It can allow a robotic to take enter knowledge generated by an skilled performing the actions to be discovered, and mix it with deep studying methods to allow simpler studying of actions. 

Much of the current work utilizing imitation and broader deep studying methods has concerned small-scale robots, and there will likely be many challenges to beat to use the identical capabilities to life-size robots, however these advances open new pathways for innovation in bettering robotic locomotion. 

Read More:  Sonantic is ready to convince listeners that synthetic voices can cry

The inspiration from animal behaviors has additionally prolonged to robotic design, with firms equivalent to Agility Robotics and Boston Dynamics incorporating drive modeling methods and integration of full-body sensors to assist their robots extra carefully mimic how animals execute complicated actions. 


Add comment