Exploring Capability-Based Control Distributions of Human-Robot Teams Through Capability Deltas: Formalization and Implications

TL;DR

This work tackles the lack of formal capability quantification in human–robot teaming by introducing Capability Deltas, a framework that measures the gap between a team’s combined capabilities and work-process requirements. Building on IMBA-derived capability profiling and a capabilities-requirement (CR) diagram, it defines a multi-dimensional capability space and a delta-based approach to adapt shared control, including the delta compensation pattern that pairs conjugated capabilities to close gaps. The approach formalizes how human and autonomous agents contribute, how resources shape capability expression, and how arbitration can be informed by observed capabilities. An exploratory study with a paraplegic participant and a collaborative robot demonstrates how capability deltas, CR diagrams, and delta compensation can explain teaming outcomes and guide adaptive control, with implications for broader domains such as manufacturing, elder care, and autonomous vehicles.

Abstract

The implicit assumption that human and autonomous agents have certain capabilities is omnipresent in modern teaming concepts. However, none formalize these capabilities in a flexible and quantifiable way. In this paper, we propose Capability Deltas, which establish a quantifiable source to craft autonomous assistance systems in which one agent takes the leader and the other the supporter role. We deduct the quantification of human capabilities based on an established assessment and documentation procedure from occupational inclusion of people with disabilities. This allows us to quantify the delta, or gap, between a team's current capability and a requirement established by a work process. The concept is then extended to the multi-dimensional capability space, which then allows to formalize compensation behavior and assess required actions by the autonomous agent.

Figures (5)

• Figure 1: Arbitration of a shared task. Flemisch et al. Flemisch.2012 extended by the agents perceiving their individual capabilities. Arrows indicate information flow.
• Figure 2: Layers of capability abstraction from real-world quantities to team capabilities.
• Figure 3: Exemplary CR diagram based on the IMBA scale ($\max(\mathbf{Q})=5$). A control distribution may be chosen in the collaborative capability space for any capability. Given a summative capability, also the summative capability space may be used. A feasible control distribution shall be chosen on the requirement line (here $f_{j}(c_{j}^{T})\equiv r_{j}^{k}=4$). Note that a $f_{j}$ similar to Equation \ref{['eq:linear_summative']} is used. In this example, the team can fulfill the requirement given any type of capability.
• Figure 4: Delta compensation pattern for two conjugated capabilities. The black dot marks the chosen control distributions. We choose distributions with highest human contribution.
• Figure 5: Exploration of manufacturing of a robot gripper, selected capabilities are most influential in the individual task. (1; from left) Participants hauls part. The left arm rests on the table, resulting in bad posture. (2) Participant positions screw. The major disability while pinch gripping influences the twisting motion. (3) Robot supports by handing over the part at pure arm's reach. (4) Robot supports by pre-positioning the screw in the hole. Participant shifts from gripping to screwing.