Hyperdimensional Computing: Robots with Memories

iClebo 07/15/2019 206



Do robots have memories?


That might seem like a question from a sci-fi movie, but with the advances in AI and robotics, it’s now a legitimate question. 


Just think of how important it can be if robots can actually remember.  That would take them one step closer to being more like humans rather than just machines.


Robots and Memories


In order for robots to become completely independent from human intervention, they need to be able to perceive the world that surrounds them and be able to navigate their surroundings.


Scientists have come up with a theory of hyperdimensional computing that will allow robots to store data regarding movement in complicated environments.


These are improvements in perception that will help AI to create their own memories.


Closing the Gap


Hearing this development might alarm some people again.


Already, there are those who are afraid that the rapid advances in AI and robotics could one day lead to the end of human civilization. They are afraid that once we create robots that are too intelligent, they would soon rise up against us in a Matrix-like scenario.


We are still far away from that.  


The present-day robots are still not fully autonomous or capable of independent thought and movement, at least not in the way that we see them in movies or on TV shows. Though we are still a long way off, scientists are fast closing the gap now.


To close the gap, scientists from the University of Maryland Institute for Advanced Computer Studies proposed what is known as hyperdimensional binary vectors (HBVs). HBVs are supposed to work in environments where there is a lot of information. The aim is to keep all the sensory information being received by the robot in one area, basically creating its memories.


A robot with its own memories will be better at making decisions on its own and expect certain situations.