Through the Clutter: Exploring the Impact of Complex Environments on the Legibility of Robot Motion

TL;DR

This work addresses the gap between legible robot motion research and real-world clutter by introducing an entropy-based clutter measure and an entropy-scaled potential field planner. The approach directly ties environmental clutter to planning through a clutter factor $\xi = e^{-D(p\|q)}$, shaping repulsive forces to promote legibility while maintaining collision avoidance. In-person validation with Baxter robots shows the entropy-scaled planner outperforms the state-of-the-art in cluttered environments and highlights that clutter dramatically affects legibility, underscoring the need for clutter-aware benchmarking and methods. The findings advance practical legible-motion planning for human-robot collaboration by demonstrating that planning in cluttered spaces requires explicit handling of environmental entropy and that results from uncluttered environments do not generalize to cluttered settings.

Abstract

The environments in which the collaboration of a robot would be the most helpful to a person are frequently uncontrolled and cluttered with many objects present. Legible robot arm motion is crucial in tasks like these in order to avoid possible collisions, improve the workflow and help ensure the safety of the person. Prior work in this area, however, focuses on solutions that are tested only in uncluttered environments and there are not many results taken from cluttered environments. In this research we present a measure for clutteredness based on an entropic measure of the environment, and a novel motion planner based on potential fields. Both our measures and the planner were tested in a cluttered environment meant to represent a more typical tool sorting task for which the person would collaborate with a robot. The in-person validation study with Baxter robots shows a significant improvement in legibility of our proposed legible motion planner compared to the current state-of-the-art legible motion planner in cluttered environments. Further, the results show a significant difference in the performance of the planners in cluttered and uncluttered environments, and the need to further explore legible motion in cluttered environments. We argue that the inconsistency of our results in cluttered environments with those obtained from uncluttered environments points out several important issues with the current research performed in the area of legible motion planners.

Figures (5)

• Figure 1: Baxter humanoid robots with the two environment setups used in the validation study.
• Figure 2: Visual explanation of the path generated by the potential field, displayed with three obstacles.
• Figure 3: Comparisons of the trajectories of the state-of-the-art legible motion planner (gray) and our proposed legible motion planner based on potential fields (blue). Our entropy-scaled potential field legible motion planner avoids other objects (green), which leads to more legible motion.
• Figure 4: In an uncluttered environment our entropy-scaled potential field legible motion planner and the state-of-the-art legible motion planner perform similarly well (p-values $>$ 0.05). In both sections of the trajectory, participants seem quite certain which object will be picked, especially when comparing the values with the values in Fig. \ref{['fig_rank']} (lower distances to the target object are better).
• Figure 5: In a cluttered environment our entropy-scaled potential field planner performed significantly better compared to the current state-of-the-art legible motion planner with a \ref{['section1']} p-value $<$ 0.0001 and a \ref{['section2']} p-value $<$ 0.05 (calculated with the Wilcoxon signed-rank test for the first rank). Especially in the first trajectory section, participants seemed uncertain regarding which object the robot would grasp (lower distances to the target object are better).