Mixed-Initiative Human-Automated Agents Teaming: Towards a Flexible Cooperation Framework

Abstract

The recent progress in robotics and artificial intelligence raises the question of the efficient artificial agents interaction with humans. For instance, artificial intelligence has achieved technical advances in perception and decision making in several domains ranging from games to a variety of operational situations, (e.g. face recognition [51] and firefighting missions [23]). Such advanced automated systems still depend on human operators as far as complex tactical, legal or ethical decisions are concerned. Usually the human is considered as an ideal agent, that is able to take control in case of automated (artificial) agent's limit range of action or even failure (e.g embedded sensor failures or low confidence in identification tasks). However, this approach needs to be revised as revealed by several critical industrial events (e.g. aviation and nuclear powerplant) that were due to conflicts between humans and complex automated system [13]. In this context, this paper reviews some of our previous works related to human-automated agents interaction driving systems. More specifically, a mixed-initiative cooperation framework that considers agents' non-deterministic actions effects and inaccuracies about the human operator state estimation. This framework has demonstrated convincing results being a promising venue for enhancing human-automated agent(s) teaming.

Mental Workload Estimation Based on Physiological Features for Pilot-UAV Teaming Applications

“…Moreover, the partial observability of the human pilot state and the low accuracy of the monitoring system (e.g., classifier) could be also compensated by such an appropriate sequential decision-making framework. Partial Observable Markov Decision Process seems to be an excellent long-term decision framework to implement such an adaptive symbiotic teaming (Chanel et al, 2020b ; Roy et al, 2020 ).…”

“…The implementation of such a MUM-T organization will require more cooperation and coordination between the agents, that could increase the mental workload of the human agent. However, there are immense advantages to this approach as for instance: benefiting from the faster and more calculative capabilities of the artificial agents, and for the human agents' better perception, judgment abilities and critical thinking (de Souza et al, 2020 ), increasing mission achievement chances while ensuring safety (Chanel et al, 2020b ), or enabling a better proximity and state awareness of the human agents (Strenzke et al, 2011 ).…”

“…But, to fine-tune the interaction capabilities, the mixed-initiative approach can be seen as a hybrid of the teamwork-centered and predictive approach of adaptive autonomy. In which, the tasks are shared between both types of agents, and the authority to lead a task is negotiated based on the state (e.g., current capabilities) of each agent (Chanel et al, 2020b ). The key idea of mixed-initiative systems is to let agents work more effectively, independently but cooperatively as a team.…”

How Can Physiological Computing Benefit Human-Robot Interaction?

“…This task reallocation, which can be roughly defined as a Mixed-Initiative Interaction (MII) [28,40,41], is particularly interesting as it will mitigate the occurrence of critical situations. MII is a promising and flexible framework that offers the possibility to integrate the notion of agents' current capabilities [42]. An MII system would allow the best current agent to seize control when necessary.…”

“…were not yet fully implemented [42]. Gombolay and collaborators studied a mixed-initiative human-robot teaming in which human factors are considered by a robot in the decision-making process [30].…”