Autonomous robots have the potential to play a critical role in urban search and rescue (USAR) by allowing human counterparts of a response team to remain in remote, stable locations while the robots execute more dangerous work in the field. However, challenges remain in developing robot capabilities suitable for teaming with humans. Communicating effectively is one of these challenges, especially if plan deviations during field operations require robot explanation. A virtual USAR team task experiment was conducted in Minecraft with a confederate acting as the remote robot. Four explanation-based communication conditions were tested: (1) always explain-the robot automatically provided explanations for any off-plan behaviors, (2) explain if asked-the robot provided an explanation only when the human counterpart requests it, (3) pull prime-the same as (2) but participants also experienced implicit training to pull information from the robot, and (4) never explain-a baseline condition in which the robot acknowledged requests but would not provide an explanation. Results indicate that the training in (3) generated more team communication than (1), but this did not improve team performance or shared situation awareness. Rather, team performance and shared situation awareness was best supported by a moderate level of explanations and the robot pushing information. These findings reinforce the importance of designing robot communication strategies that can reduce human workload, particularly communication overhead, in dynamic and time-constrained tasks.
This study focuses on methodological adaptations and considerations for remote research on Human-AI-Robot Teaming (HART) amidst the COVID-19 pandemic. Themes and effective remote research methods were explored. Central issues in remote research were identified, such as challenges in attending to participants' experiences, coordinating experimenter teams remotely, and protecting privacy and confidentiality. Instances of experimental design overcoming these challenges were identified in methods for recruitment and onboarding, training, team task scenarios, and measurement. Three case studies are presented in which interactive in-person testbeds for HART were rapidly redesigned to function remotely. Although COVID-19 may have temporarily constrained experimental design, future HART studies may adopt remote research methods to expand the research toolkit.
Artificial social intelligence (ASI) agents have great potential to aid the success of individuals, human–human teams, and human–artificial intelligence teams. To develop helpful ASI agents, we created an urban search and rescue task environment in Minecraft to evaluate ASI agents’ ability to infer participants’ knowledge training conditions and predict participants’ next victim type to be rescued. We evaluated ASI agents’ capabilities in three ways: (a) comparison to ground truth—the actual knowledge training condition and participant actions; (b) comparison among different ASI agents; and (c) comparison to a human observer criterion, whose accuracy served as a reference point. The human observers and the ASI agents used video data and timestamped event messages from the testbed, respectively, to make inferences about the same participants and topic (knowledge training condition) and the same instances of participant actions (rescue of victims). Overall, ASI agents performed better than human observers in inferring knowledge training conditions and predicting actions. Refining the human criterion can guide the design and evaluation of ASI agents for complex task environments and team composition.
To support research on artificial social intelligence for successful teams (ASIST), an urban search and rescue task (USAR) was simulated within Minecraft to serve as a Synthetic Task Environment (STE). The goal for the development of the present STE was to create an environment that provides ample opportunities to allow ASI agents to demonstrate the theory of mind by making inferences and predictions of humans’ states and actions in the USAR task environment, and in the future to intervene to improve teamwork in real-time. This paper describes the STE design background, design potentials and considerations, rich data collection opportunities, and potential usage for more broad research.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.