The giant social media company is now trying to make the robots smarter as they were ever before. A blog post has been released (along with a video) by the company showing that it is training the robots to handle multiple situations without losing their functionalities. In the video, the robot is going through the vertical hills, murky grass, and even from sand ruses. The trainer is also throwing some payloads on robots to check whether they can handle such load without losing their foot or not.
The latest machine learning model that has been developed by Facebook in collaboration with RMA and UC Berkeley is trying to give the robot such power to quickly respond to the sudden variations in the conditions. However, there is no doubt that it is very difficult to get used to each and every situation through pre-programming as there can be countless situations that can be happened in the real life unexpectedly. Still, the robot is learning to get used to the sudden situations that were unknown before. The trainers are passing robots through enormous trials to learn as the newborn baby learns.
The latest machine learning model that has been developed by Facebook in collaboration with RMA and UC Berkeley is trying to give the robot such power to quickly respond to the sudden variations in the conditions. However, there is no doubt that it is very difficult to get used to each and every situation through pre-programming as there can be countless situations that can be happened in the real life unexpectedly. Still, the robot is learning to get used to the sudden situations that were unknown before. The trainers are passing robots through enormous trials to learn as the newborn baby learns.
The director of Facebook AI said that the company wants robots to learn to survive the abrupt conditions. The teams of experts are trying to include computer visualization. The robotic lab of the company is developing robots to contemplate themselves as the company is teaching robots to build a common intelligence for grasping multifaceted tasks in unaware situations. We all know that human beings and even animals learn different things through their observation and it is very difficult to create a sense in the robot. The technology of the company has moved into a new age where these robots will learn how the ecosphere works and how to execute different tasks through their intelligence. The company is further planning to create such agents to help people in their house works during the daily life.
Facebook has also created a curated collection of 1000 3D realistic image spaces obtained from Matterport where robots will be able to recognize different stuff so that they can fetch such objects when they are ordered. The company’s AL expert said that this openly available data of 3D photos are meant to teach robots to identify free spaces and to spot stuff placed. The social media company further said that it has about five million dollhouses models of building structure that the company utilizes for numerous drives such as house visits or fixtures enactment and it will also use in the training of robots.
Facebook has also created a curated collection of 1000 3D realistic image spaces obtained from Matterport where robots will be able to recognize different stuff so that they can fetch such objects when they are ordered. The company’s AL expert said that this openly available data of 3D photos are meant to teach robots to identify free spaces and to spot stuff placed. The social media company further said that it has about five million dollhouses models of building structure that the company utilizes for numerous drives such as house visits or fixtures enactment and it will also use in the training of robots.
