Editorial: Variable autonomy for human-robot teaming

摘要

In the modern era, the integration of robots has become a cornerstone of progress in various sectors, 5 from entertainment and companionship to first responders and defence. One of the critical aspects of this 6 integration is the concept of Human-Robot Teaming (HRT). Unlike traditional automation, where robots 7 operate in isolation or perform predefined tasks, HRT involves robots working alongside humans and 8 aiming to achieve shared goals, often in dynamic and interactive environments. This collaboration leverages 9 the strengths of both humans and robots to achieve goals that neither could accomplish as effectively alone.10 Such capabilities often involve teaming with well-defined and static control frameworks: humans in-, 11 on-, or out-of-theloop. However, as the world is not static and more complex tasks are required of the 12 team, the ability to dynamically allocate tasks with varying levels of autonomy becomes essential. Variable 13 Autonomy (VA) refers to the ability of the robotic systems to dynamically vary their level or degree of Human-robot teams must handle a diverse array of tasks of variable complexity, from gross and fine 25 motor skills to visual perception, cognitive processing, and speech. Such tasks may coincide or occur in 26 quick succession. For efficient teaming, robots must be able to identify these composite, concurrent tasks 27 performed by humans.28 (Baskaran and Adams, 2023) review over a hundred task recognition algorithms and evaluate them on six 29 criteria: sensitivity, suitability, generalizability, composite factor, concurrency, and anomaly awareness.Through the extensive review, (Baskaran and Adams, 2023) make multiple recommendations for future 31 directions, including the need for ecologically valid HRT datasets and adaptively segmenting metrics.The need for more efficient metrics in HRT is tackled by another paper of our collection. (Verhagen 33 et al., 2024) propose an evaluation method to verify if dynamic task allocation using variable autonomy in 34 human-robot teams ensures not just completion of the task but also meaningful human control by satisfying 35 accountability, responsibility, and transparency. This approach quantifies traceability both subjectively and 36 objectively by analysing human responses during and after simulated collaborative activities. Additionally, 37 it incorporates semi-structured interviews following the simulation to uncover the underlying reasons 38 for the outcomes and gather suggestions for enhancing the variable autonomy strategy. In their article, a 39 real-world illustration with firefighters is presented.ROBOT PERCEPTION AND AUTONOMY ADJUSTMENT (Lakhnati et al., 2024) the use of Large Language Models (LLMs) to facilitate variable autonomy through