Design and develop algorithms for safe navigation of robots, especially social robots, in the presence of humans.
Mankind’s long-held dream is to build humanoid robots which interact well with people. Every day, human beings are completely subconsciously choosing their path of movement while noticing so many aspects, such as avoiding collisions with objects or other people, crossing smooth paths and achieving the desired goal, as well as observing many unwritten customary laws; yet robots do not know any of these rules.
It is desirable to see a human-like movement from robots, especially in the areas where people are present. The first step in imitating how humans move and behave is to identify and understand these behaviors. Vision is the most important sensation humans rely on to understand the environment, and for this reason, this sense has been the most studied in machine learning and artificial intelligence.
One of the most interesting areas in machine vision and robotics is the design of learning algorithms which have high accuracy, low computational complexity and enable autonomous mobile robots to operate in interactive visual environments. One of the most powerful tools available to researchers for this purpose is deep learning. The use of deep learning methods has been exploited in many real-world issues nowadays, and it is hoped that it will respond appropriately to this issue as well.