Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Control-Theoretic Analysis of Shared Control Systems

Reuben M. Aronson, Elaine Schaertl Short

TL;DR

This work reframes shared-control systems as user-controlled dynamical plants, enabling evaluation of user experience without strict assumptions about user behavior. It analyzes the popular shared-autonomy algorithm as a dynamical system that behaves like an integrator, revealing a runaway goal-confidence issue and a fundamental restriction that the system imposes on possible user actions. By deriving a two-phase dynamic—outside the convex hull of goals the system drags toward the hull, and inside the hull the user influences a set-point via goal probabilities—the paper explains observed user satisfaction dynamics and equilibria. The proposed control-theoretic lens offers design guidance for shared-control systems, including stabilization of goal updates and understanding the trade-offs between automation and user control, with potential applicability to a broad class of assistive technologies.

Abstract

Users of shared control systems change their behavior in the presence of assistance, which conflicts with assumpts about user behavior that some assistance methods make. In this paper, we propose an analysis technique to evaluate the user's experience with the assistive systems that bypasses required assumptions: we model the assistance as a dynamical system that can be analyzed using control theory techniques. We analyze the shared autonomy assistance algorithm and make several observations: we identify a problem with runaway goal confidence and propose a system adjustment to mitigate it, we demonstrate that the system inherently limits the possible actions available to the user, and we show that in a simplified setting, the effect of the assistance is to drive the system to the convex hull of the goals and, once there, add a layer of indirection between the user control and the system behavior. We conclude by discussing the possible uses of this analysis for the field.

Control-Theoretic Analysis of Shared Control Systems

TL;DR

This work reframes shared-control systems as user-controlled dynamical plants, enabling evaluation of user experience without strict assumptions about user behavior. It analyzes the popular shared-autonomy algorithm as a dynamical system that behaves like an integrator, revealing a runaway goal-confidence issue and a fundamental restriction that the system imposes on possible user actions. By deriving a two-phase dynamic—outside the convex hull of goals the system drags toward the hull, and inside the hull the user influences a set-point via goal probabilities—the paper explains observed user satisfaction dynamics and equilibria. The proposed control-theoretic lens offers design guidance for shared-control systems, including stabilization of goal updates and understanding the trade-offs between automation and user control, with potential applicability to a broad class of assistive technologies.

Abstract

Users of shared control systems change their behavior in the presence of assistance, which conflicts with assumpts about user behavior that some assistance methods make. In this paper, we propose an analysis technique to evaluate the user's experience with the assistive systems that bypasses required assumptions: we model the assistance as a dynamical system that can be analyzed using control theory techniques. We analyze the shared autonomy assistance algorithm and make several observations: we identify a problem with runaway goal confidence and propose a system adjustment to mitigate it, we demonstrate that the system inherently limits the possible actions available to the user, and we show that in a simplified setting, the effect of the assistance is to drive the system to the convex hull of the goals and, once there, add a layer of indirection between the user control and the system behavior. We conclude by discussing the possible uses of this analysis for the field.
Paper Structure (10 sections, 12 equations, 1 figure)

This paper contains 10 sections, 12 equations, 1 figure.

Table of Contents

  1. Introduction
  2. Background
  3. Shared Control
  4. Control Theory
  5. Dynamical Analysis of Shared Autonomy
  6. Background: Shared Autonomy
  7. Dynamical Systems Representation
  8. Dynamics Analysis
  9. Experience of User Control
  10. Discussion and Conclusion

Figures (1)

  • Figure 1: While there may be many possible actions that the system can take in some state $x$, here plotted with their Q-values corresponding to two different goals, the assistance system only ever selects actions that lie on the Pareto frontier, indicated by the blue circles.