Rish is an entrepreneur and investor. Previously, he was VC at Gradient Ventures (Google's AI fund), co-founder of a fintech startup that built an analysis platform for SEC filings, and worked as a PhD student in computer science at MIT on deep learning research.
Other contributions from this contributor
- Assessment of the potential for a gig economy in education
- Generative algorithms redefine the interface between software and music
In the past For two decades, humanoid robots have significantly improved their ability to perform functions such as capturing objects and using computer vision to recognize things since the release of Honda ASIMO Robots in 2000. Despite these improvements, their ability to walk, jump, and perform other complex leg movements as smoothly as humans remained a challenge for robotics.
In recent years, new advances in robot learning and design have been using animal behavior data and insights to enable legged robots to move much more humanly.
Researchers from Google and UC Berkeley published work Earlier this year, this showed a robot how to walk by mimicking a dog's movements using a technique called imitation learning. Separate job showed a robot that successfully learned to walk by trial and error using learning algorithms for depth enhancement.
In particular, imitation learning was used in robotics for various applications such as OpenAIs job It helps a robot capture objects by imitation, but its use in robot movement is new and encouraging. It can allow a robot to collect input data generated by an expert performing the actions to be learned and combine it with deep learning techniques to enable more effective learning of movements.
Much of the recent work on counterfeiting and more extensive deep learning techniques has involved small robots, and there will be many challenges to be met in order to apply the same skills to life-size robots. However, these advances open up new avenues for innovation in improving robot motion.
The inspiration from animal behavior has also spread to robot design, with companies like Agility robotics and Boston Dynamics Integration of force modeling techniques and integration of whole body sensors to help their robots more accurately mimic the execution of complex movements by animals.