There a number of choices which a human being has to make in a day. Especially, the choices about where to visit and how to reach there. Researchers from MIT are trying to make robots go through the same choice in navigation. With a new-motion planning model and neural network, robots will be able to find how to get to their end goal after learning more about the environment. The algorithm helps by making a ‘tree’ of possible decisions which keeps on branching until the robot finds the most appropriate path to navigation.
This implies that in the future, robots will take us where we want to reach and quicker without using much computing power. Old algorithms were unable to make robots learn from its mistakes. Also, the robots had no idea how they acted previously in the similar environment. This is what the researchers from MIT wanted to fix. With the old algorithm, even after a number of times, the robots were as confused about the environment as they were at the first time this is because they were always exploring and never observing. The neural network allows the system to guide the robot through complicated environments and then help the robot to apply the same strategies in other situations.
They put the robots in situations where they had to find their way out of an inner chamber through a narrow channel. In the next situation, they had to face moving elements just like traffic or a crowded street. The researchers used a roundabout, it is the situation in which there is a reasoning about how others will respond to your actions and then how you respond to their actions. The model learned how to handle what other vehicles were doing. It is amazing as well as terrifying at the same time that robots are learning this high powered navigation.
Searching for new things, she has found herself as a writer. In the midst of day to day routine life, she finds her peace in reading and painting. With a passion for reading, she believes in learning new things to add value to her own as well as to the lives of others around her.