Project Aquaticus

Project Aquaticus

Project Aquaticus@West Point

Project Aquaticus is a human-robot teaming testbed designed to explore humans and robots being partners while embedded in the same environment. Project Aquaticus embeds humans and robots as teammates playing games of capture the flag against similarly situated human-robot teams. The research goal of the testbed is to explore the interactions of human-robot trust, human cognitive load, and robot autonomy. Project Aquaticus was established at the Massachusetts Institute of Technology in 2015. In 2019, the Project Aquaticus testbed migrated to the Robotics Research Center at the United States Military Academy at West Point.

Usma Crest
Trendy Pants and Shoes
Usma Crest

Human-robot Teaming

Human-robot teaming has traditionally been studied in sterile and very controlled environments such as industrial settings or in a laboratory.  Project Aquaticus breaks out of that paradigm to transition humans and robots to be embedded in the same environment as teammates.  The environment is made more exciting by having human-robot teams play against each other in games of capture the flag. Project Aquaticus explores human-robot interaction in the forms of multi-modal communication, authority and team dynamics, human training, and robot autonomy.

Usma Crest

AR/VR

Project Aquaticus leverages Augmented and Virtual Reality to study various facets of human-robot teaming.  In general, Project Aquaticus studies the human's cognitive load and searches for ways in which to reduce that load.  One method of addressing the reduction of cognitive load is to research multi-modal interactions between humans and robots.  Augmented Reality is one such modality that Project Aquaticus is exploring for improved human-robot teaming communication.  A major collaboration with the Naval Postgraduate School (NPS) is addressing the training of humans for human-robot teams and their learning of tactics.  Virtual Reality allows us to perform repeated experiments while reducing the overall cost (in time and hardware) required from performing these experiments in the real world.

Trendy Pants and Shoes
Usma Crest
Usma Crest

Machine Learning with pLearn

Project Aquaticus supports machine learning for individual robot behaviors and at the strategic level through the use of pLearn.  moos-ivp-pLearn incorporates a popular machine learning library TensorFlow/Keras with MOOS-IvP and Project Aquaticus capture the flag simulation.  Such an integration allows researchers to explore the current implementation with Deep Reinforcement Learning.  Students using moos-ivp-pLearn explore changing the learned model parameters and the reward function for various behaviors in capture the flag.  Additionally, researchers can use this example integration to integrate their own machine learning algorithms with Project Aquaticus and MOOS-IvP.

Trendy Pants and Shoes
Usma Crest