Teammate-pattern-aware autonomy based on organizational self-design principles
- PDF / 2,777,073 Bytes
- 30 Pages / 439.37 x 666.142 pts Page_size
- 88 Downloads / 167 Views
(2020) 34:39
Teammate‑pattern‑aware autonomy based on organizational self‑design principles Edmund H. Durfee1 · Abhishek Thakur1 · Eli Goldweber1
© Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract We describe an approach for constraining robot autonomy based on the robot’s awareness of patterns of its human teammates’ behaviors, rather than either ignoring its teammates (which is fast but dangerous) or inferring their plans (which is safer but slow). We explore the promise, and limitations, of this approach in a series of simulated problems where an unmanned ground vehicle and its human teammates must rapidly respond to a sudden context shift. Our results help us discern conditions under which a pattern-aware approach can be more effective than the alternatives, and our current efforts investigate how the manned– unmanned team can adopt biases to more readily establish such conditions that are more favorable to the pattern-aware approach. Keywords Human–robot coordination · Autonomy · Organizational design · Multiagent planning
1 Introduction Our research is concerned with mixed manned–unmanned teams (specifically involving unmanned ground vehicles (UGVs)) undergoing sudden context shifts, such as when a natural disaster strikes or an adversary attacks. Under nominal mission conditions, a human operator can closely oversee a UGV, but during sudden context shifts the operator must often focus on personally-important tasks like self preservation, and thus at such times can least afford to provide guidance/supervision to the UGV. As a consequence, it is at precisely these times of change and uncertainty that the UGV should shoulder greater autonomy for controlling its own responses. The danger, though, is that a UGV’s autonomous Distribution A: Approved for public release; distribution unlimited. OPSEC # 2970. * Edmund H. Durfee [email protected] Abhishek Thakur [email protected] Eli Goldweber [email protected] 1
Computer Science and Engineering, University of Michigan, Ann Arbor, USA
13
Vol.:(0123456789)
39
Page 2 of 30
Autonomous Agents and Multi-Agent Systems
(2020) 34:39
behavior, based on algorithmic assessments balancing what it knows and has learned about the risks and rewards of different courses of action in its perceived environment, might deviate from its human teammates’ immediate expectations, compounding their confusion, and threatening team goals and even teammate safety [14]. Hence, robotic agents that can flexibly come into contact with human teammates [22] may need to trade away some degree of task-execution optimality in order to satisfy teammates’ preferences [15]. This problem, of how a UGV can rapidly exercise its autonomy in ways that support, not confound, its human teammates, is the focus of this article. We characterize teammate-aware autonomy as when a UGV’s autonomous behaviors are informed and constrained by awareness of its teammates’ plans and expectations. Teammate-aware autonomy can be (and has been) realized in various ways. For example
Data Loading...
