Now, It Is Almost Possible For Robots To Learn New Things On Their Own And Impart The Knowledge On Other Robots
Already, robots have taken over several human activities like packaging in warehouses, positioning products in line of production, assisting bedridden patients and several activities. But robots cannot still handle certain tasks because of the limitation in their knowledge. They still find it difficult to recognize and figure out the best ways to handle many other objects.
During childhood, human beings have been able to build a vast data collection of several objects and how to handle them. This is why people know how to handle a plastic bowl, glass cup, a raw egg and a live grenade or some other explosive devices. Stefanie Tellex, a professor of computer science at Brown University has already started working on feeding contemporary robots with all the required knowledge to bridge the gap.
For robots to be able to identify, grasp and handle different types of objects, they need to be fed with reams of data. This data will definitely come from long rigorous programming consisting of several thousand lines of codes. To replicate the knowledge on millions of robots worldwide, the knowledge should be deposited on a central platform where other robots can access the knowledge.
This is the main purpose of Professor Tellex’s project which has been tagged “Million Object Challenge”. In other words, the project is for robots around the world to learn these activities, upload the knowledge data to the cloud and give access to the data to other robots for analyses and usage of the data.
As a proof of conversance with Robots, there is a robot in Professor Telex’s lab. This machine was produced by Rethink robotics. This robot has the ability to scan small hairbrushes. Not only that, it has the ability of taking series of pictures with a camera and also measuring certain depth of some blocks with an infrared sensor.
It later picked up the hairbrush after several attempts and also shook the brush very well to see if it will drop but it didn’t. Apart from this particular robot, only very few other robots can perform those tasks- if any. It shows that this robot has acquired one particular knowledge that puts it a step ahead of other robots.
Won’t it be amazing if this robot can impart this knowledge to other robots through a centralized platform for robots? It will be simply mind blowing if robots are able to pick up almost all objects like human beings do. This is why Professor Tellex and one of her students have started compiling data on how robots can pick up several objects like a toddler’s sippy cup, shoe, plastic boat, rubber duck and several other things. In fact, she already has about 200 objects on her list.
In conclusion, she has left the project open for other scientists to contribute. There is nothing wrong in robots having the ability to identify and pick up over one million objects. Not only that, robots should also be able to search through a crowded shelf and pick out one particular item.