Cost Functions for Robot Motion Style

TL;DR

This paper explores cost function representations for generating expressive robot motion, comparing feature-based and neural network approaches, and evaluates their effectiveness through user studies across various styles.

Contribution

It introduces two methods for encoding style in robot motion via cost functions and compares their performance and advantages in a trajectory optimization framework.

Findings

Both approaches perform similarly and outperform the baseline.

Feature-based costs require fewer parameters and may excel on some styles.

Neural network costs can learn more complex, nuanced styles without expert-designed features.

Abstract

We focus on autonomously generating robot motion for day to day physical tasks that is expressive of a certain style or emotion. Because we seek generalization across task instances and task types, we propose to capture style via cost functions that the robot can use to augment its nominal task cost and task constraints in a trajectory optimization process. We compare two approaches to representing such cost functions: a weighted linear combination of hand-designed features, and a neural network parameterization operating on raw trajectory input. For each cost type, we learn weights for each style from user feedback. We contrast these approaches to a nominal motion across different tasks and for different styles in a user study, and find that they both perform on par with each other, and significantly outperform the baseline. Each approach has its advantages: featurized costs require learning fewer parameters and can perform better on some styles, but neural network representations do not require expert knowledge to design features and could even learn more complex, nuanced costs than an expert can easily design.