Human-in-the-Loop Optimization of Shared Autonomy in Assistive Robotics

IEEE Robot Autom Lett, 2017 · DOI: 10.1109/LRA.2016.2593928 · Published: January 1, 2017

Simple Explanation

This paper introduces a new way to customize how robots help people with motor impairments. Instead of relying on computer algorithms, the user gets to fine-tune the robot's assistance based on their own preferences. The approach was tested with a robotic arm and a group of people, some with spinal cord injuries. The results showed that people could successfully adjust the robot's behavior to suit their needs. Interestingly, people didn't always choose the settings that made them perform the best. Some preferred to have more control, even if it meant the task took longer or was more difficult. This highlights the importance of letting users customize their assistive technology.

Study Duration

Not specified

Participants

17 subjects (4 with spinal cord injury, 13 without injury)

Evidence Level

Level Not specified, Pilot study

Key Findings

1
All subjects were able to converge to an assistance paradigm, suggesting the existence of optimal solutions for user-driven customization of shared autonomy in assistive robotics.
2
The amount of assistance was not always optimized for task performance, with some subjects favoring retaining more control even at the expense of performance.
3
Customized assistance led to statistically equivalent performance between SCI and uninjured subjects, reducing performance differences.

Research Summary

This paper proposes a framework for user-driven customization of shared autonomy in assistive robotics, formulating it as a nonlinear optimization problem where users tune system parameters. A pilot study with 17 subjects (4 with spinal cord injury, 13 without injury) explored interactive optimization with a robotic arm, showing that users could converge to an assistance paradigm. The study supports user-driven customization, revealing that users may prioritize control over pure task performance, and provides guidance for continued development and study in this area.

Practical Implications

Personalized Assistive Technology

The research highlights the importance of allowing users to customize their assistive devices to match their individual needs and preferences, potentially leading to greater adoption and satisfaction.

Complex Cost Functions

The findings suggest that users consider factors beyond simple performance metrics (e.g., comfort, sense of control) when interacting with assistive robots, indicating that cost functions in robot control need to account for these subjective elements.

Rehabilitation Strategies

The study provides insights for tailoring rehabilitation programs by understanding how individuals with motor impairments interact with assistive technologies and how their preferences evolve over time.

Study Limitations

1
The exact nature of the cost function optimized by users during customization remains indeterminate.
2
The interactive optimization procedure was not fully automated, relying on a system operator to translate verbal commands.
3
The pilot study involved a limited number of participants, especially those with spinal cord injuries.

Your Feedback

Was this summary helpful?

Back to Assistive Technology