ARL's Intelligent Systems Enterprise vision is to enable the teaming of autonomous intelligent systems with soldiers in dynamic, unstructured combat environments, as well as in non-combat military installations and base operations. To accomplish this vision for interdependent soldier-robot teaming, there has been a paradigm shift in robotic research conducted by ARL from the current instantiation of fielded remote-controlled or teleoperated robots to systems with increased intelligence, decision-making capability, and autonomy. This type of teaming is needed for future joint, interdependent, network-enabled operations.
While the technological capabilities of robotic systems are advancing by the day, many of these new systems are still in the early stages of research and development. In many cases, areas of need or potential use have been identified, preliminary requirements have been created, and engineering solutions for prototype systems have been researched. However, the human element must be considered early on in this design process, because without considering human factors, such as human-system interfaces, performance, as well as trust and expectation, the potential result will be limited or inappropriate use of the system.
Computer-based simulation provides a safe, economical, and efficient means to assess human-robot interaction (HRI) throughout the robot life cycle. Simulation provides the opportunity for collaboration between soldiers and unmanned platforms at multiple levels in the design process, which can help guide the design process and provide needed feedback for development.
One such tool useful for analyzing HRI is the Robotic Interactive Visualization and Experimentation Technology (RIVET) simulation environment. RIVET version 1.0 was built using the Torque Game Engine (TGE) version 1.5.2, designed by GarageGames. It was designed by the ARL's Robotics Collaborative Technology Alliance (RCTA) to allow engineers to test and debug intelligence and perception algorithms for autonomous unmanned vehicles prior to field exercises. Simulation tools like RIVET can provide an essential initial evaluation of robotic algorithms and concepts in simulated environments that can be varied systematically. While RIVET was originally designed to test and debug algorithms, the design and capabilities of the software lend themselves to the development of HRI-specific missions with unmanned ground vehicles (UGVs).
Like most commercial off-the-shelf game engines, the TGE provides functionality for graphics, physics, artificial intelligence, lighting, and many other features. It provides all the necessary attributes and core functionality to conduct virtual experiments. Building a simulated environment involves several different activities necessary for HRI experimentation: 1) creating the virtual terrain surface, 2) adding static features typically found in the scene, 3) adding dynamic elements, such as people and vehicles, 4) producing the interactive graphical user interfaces (GUIs), 5) choosing the unmanned platform for the exercise, 6) implementing a user interface for the UGV, and finally, 7) assessing the user during a study.
RIVET can be used in single-user mode or multi-user mode depending on the problem statement and research goals. In single-user mode, RIVET loads the scene and then a user is able to access the world editor, GUI editor, or control an avatar (soldier, vehicle, etc.) using keyboard or joystick. Multi-user mode allows up to 64 local area networked users to join the mission as clients. These clients can be another vehicle, robot, soldier, or a sensor that is attached to an entity in the simulation. Currently, the choice of additional vehicle sensors within RIVET are cameras, laser detection and ranging (LADAR), and a custom interface used to connect other applications to the simulation environment.
Through the RIVET main menu, there are customization options that are available through buttons at the bottom of the menu screen. It is possible to load a customized user mission (Server computer), join a mission (Client computer), or mount a sensor to a vehicle or robot on the RIVET server computer (Client computer). In addition, the configuration menu allows customization of the video screen size, resolution, and vehicles.
This work was done by Kristin E Schaefer of Oak Ridge Associated Universities, and Ralph Brewer of the Army Research Laboratory. ARL-0197
This Brief includes a Technical Support Package (TSP).
A Guide for Developing Human-Robot Interaction Experiments in the Robotic Interactive Visualization and Experimentation Technology (RIVET) Simulation
(reference ARL-0197) is currently available for download from the TSP library.
Don't have an account? Sign up here.