Jack Maxwell Vice

Analysis of Human-robot Interaction at the DARPA Robotics Challenge Trials

In December 2013, the Defense Advanced Research Projects Agency DARPA Robotics Challenge DRC Trials were held in Homestead, Florida. The DRC Trials were designed to test the capabilities of humanoid robots in disaster response scenarios with degraded communications. Each team created their own interaction method to control their robot, either the Boston Dynamics Atlas robot or a robot built by the team itself. Of the 15 competing teams, eight participated in our study of human-robot interaction. We observed the participating teams from the field with the robot and in the control room with the operators, noting many performance metrics, such as critical incidents and utterances, and categorizing their interaction methods according to the number of operators, control methods, and amount of interaction. We decomposed each task into a series of subtasks, different from the DRC Trials official subtasks for points, to gain a better understanding of each teams performance in varying complexities of mobility and manipulation. Each teams interaction methods have been compared to their performance, and correlations have been analyzed to understand why some teams ranked higher than others. We discuss lessons learned from this study, and we have found in general that the guidelines for human-robot interaction for unmanned ground vehicles still hold true: more sensor fusion, fewer operators, and more automation lead to better performance.

Defense Automated Neurobehavioral Assessment (DANA)-psychometric properties of a new field-deployable neurocognitive assessment tool.

The Defense Automated Neurobehavioral Assessment (DANA) is a new neurocognitive assessment tool that includes a library of standardized cognitive and psychological assessments, with three versions that range from a brief 5-minute screen to a 45-minute complete assessment. DANA is written using the Android open-source operating system and is suitable for multiple mobile platforms. This article presents testing of DANA by 224 active duty U.S. service members in five operationally relevant environments (desert, jungle, mountain, arctic, and shipboard). DANA was found to be a reliable instrument and compared favorably to other computer-based neurocognitive assessments. Implications for using DANA in far-forward military settings are discussed.

Simulation fidelity design informed by physiologically-based measurement tools

Virtual environments (VEs) and simulations are being employed for training applications in a wide variety of disciplines, both military and civilian. The common assumption is that the more realistic the VE, the better the transfer of training to real world tasks. However, some aspects of task content and fidelity may result in stronger transfer of training than even the most high fidelity simulations. A physiologically-based system capable of dynamically detecting changes in operator behavior and physiology throughout a VE experience and comparing those changes to operator behavior and physiology in real-world tasks, could potentially determine which aspects of VE fidelity will have the highest impact on transfer of training. Thus, development of training assessment and guidance tools that utilize operator behavior and physiology to determine VE effectiveness and transfer of training are needed.

Use of neurophysiological metrics within a real and virtual perceptual skills task to determine optimal simulation fidelity requirements

The military is increasingly looking to virtual environment (VE) developers and cognitive scientists to provide virtual training platforms to support optimal training effectiveness within significant time and cost constraints. However, current methods for determining the most effective levels of fidelity in these environments are limited. Neurophysiological metrics may provide a means for objectively assessing the impact of fidelity variations on training. The current experiment compared neurophysiological and performance data for a real-world perceptual discrimination task as well as a similarlystructured VE training task under systematically varied fidelity conditions. Visual discrimination and classification was required between two militarilyrelevant (M-16 and AK-47 rifle), and one neutral (umbrella) stimuli, viewed through a real and virtual Night Vision Device. Significant differences were found for task condition (real world versus virtual, as well as visual stimulus parameters within each condition), within both the performance and physiological data.

Perceptually-Informed Virtual Environment (PerceiVE) Design Tool

Virtual environments (VEs) are becoming more and more prevalent as training tools for both military and civilian applications. The common assumption is that the more realistic the VE, the better the transfer of training to real world tasks. However, some aspects of task content and fidelity may result in stronger transfer of training than even the most high fidelity simulations. This research effort seeks to demonstrate the technical feasibility of a Perceptually-informed Virtual Environment (PerceiVE) Design Tool, capable of dynamically detecting changes in operator behavior and physiology throughout a VE experience and comparing those changes to operator behavior and physiology in real-world tasks. This approach could potentially determine which aspects of VE fidelity will have the highest impact on transfer of training. A preliminary study was conducted in which psychophysiological and performance data were compared for a visual search tasks with low and high fidelity conditions. While no significant performance effects were found across conditions, event-related potential (ERP) data revealed significant differences between the low and high fidelity stimulus conditions. These results suggest that psychophysiological measures may provide a more sensitive and objective measure for determining VE fidelity requirements.

An Intuitive Wearable Concept for Robotic Control

In this study we explore the concept of gesture-based robot control for maneuver and manipulation, using a prototype system by AnthroTronix [1]. For the task, 24 Soldier-participants were asked to tele-operate the robot through a course containing several tight turns and obstacles. They were then asked to simulate “planting a breaching charge” by approaching a target with a marker attached to the end of the manipulator arm. They were provided with video feedback via a camera mounted to the chassis of the robot. Performance on the task was defined as time to navigate to the intended target, time to manipulate the arm to the target, and accuracy of the manipulation task. Results suggested that the use of the instrumented glove reduced the time needed to maneuver the manipulator arm as compared to the use of the handheld controller.

Augmented Interaction: Applying the Principles of Augmented Cognition to Human-Technology and Human-Human Interactions

The field of Augmented Cognition (AugCog) has evolved over the past decade from its origins in the Defense Advanced Research Projects Agency (DARPA)-funded research program, emphasizing modulation of closed-loop human-computer interactions within operational environments, to address a broader scope of domains, contexts, and science and technology (S&T) challenges. Among these are challenges related to the underlying theoretical and empirical research questions, as well as the application of advances in the field within contexts such as training and education. This paper summarizes a series of ongoing research and development (R&D) efforts aimed at applying an AugCog-inspired framework to enhance both human-technology and human-human interactions within a variety of training and operational domains.

Development of Tactile and Gestural Displays for Navigation, Communication, and Robotic Control

Cognitive demands on dismounted Soldiers are ever increasing. This is an investigation into using gestural controls and a tactile display vest to reduce cognitive, physical, and temporal demands as relevant to covert communications and robot control. Data was collected on 31 Soldiers for this experiment-based evaluation consisting of task demands of a typical rural reconnaissance patrol deconstructed into specific tasks involving a platoon leader role, a squad leader role, and a robot controller role. Results found that use of an instrumented glove and a tactile display vest was associated with a higher average percentage and faster average speed of signal detections when compared to traditional hand and arm signals. Glove-based robot control was also compared with traditional methods and evaluated.