Electronic Thesis/Dissertation
 

Biomimicking Natural Behavior Learning for an Animal-Like Companion Robot

Open Access

With people becoming increasingly available on social media nowadays, they are growing distant from each other in reality. This, along with global changes happening like climate change, depletion of fuels, etc., is increasing the amount of stress that humans, in general, are facing. All the above reasons have led to the development of new and creative ways of therapy to assist people, children, and humans in general. Recent advances in treatment have found that Animal Assisted Therapy (AAT) alleviates stress, anxiety, depression, and fatigue in humans and is especially useful for people undergoing cancer treatments, people with anxiety, people in long-term care facilities, and people with dementia, children with autism, etc. Research shows that over 50000 dogs in the US today are used as comfort dogs for Animal-Assisted Therapy.My project is created with the aim of “Everyone can own a pet.” While the actual dog is proven to be very useful in therapy, they cannot perform complex behavior / assist humans in emergencies. My project aims to create a Dog robot that not only looks realistic but would also be proven helpful in emergencies. My research involves designing and developing a hardware prototype of a quadruped robot and making it socially interactable with humans. With current advancements in machine learning and AI, we can simulate and deploy more complicated control systems in robots and teach robots to take action independently. My project involves using a reinforcement learning algorithm, which helps replicate behaviors from the actual dog. My robot uses an advanced vision system that recognizes different poses of humans and takes corresponding actions based on the identified gesture. The robot is coded in the ROS environment as it helps connect software to hardware in a simplified way. Electric circuitry was designed based on the tolerances of servo motors and the ability to power multiple systems.

Author Language Keyword Date created Type of Work License
  • All rights reserved
Rights statement GW Unit Degree Advisor Committee Member(s) Persistent URL

Notice to Authors

If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.

Thumbnail Title Date Uploaded Visibility Actions
Preview of Balaji_gwu_0075M_16075.pdf Balaji_gwu_0075M_16075.pdf 2022-10-04 Open Access