Legibot: Generating Legible Motions for Service Robots Using Cost-Based Local Planners

TL;DR

Legibot addresses legibility in service-robot navigation by introducing a legibility-aware local planner that predicts observer expectations over multiple potential goals and optimizes trajectories to resemble the true goal $G^*$ while diverging from unintended goals. The method augments the task-cost with a similarity term $C_{ ext{Sim}}$ and a field-of-view term $C_{ ext{FOV}}$, integrated as $C_{ ext{LA}}( abla)= C_{ ext{Task}}( abla) + \lambda_{ ext{sim}} C_{ ext{Sim}}( abla) + \lambda_{ ext{fov}} C_{ ext{FOV}}( abla)$, and incorporates observer perspective through $h(q)$ and $v(q)$ in the similarity metric. The contributions include (i) a novel legibility-aware planning objective, (ii) observer-perspective adaptations for partial field-of-view constraints, and (iii) a ROS-based navigation stack enabling real-time deployment on service robots, validated in Gazebo restaurant scenarios, offline user studies, and real Pepper experiments. Results indicate improved observer ability to infer the robot's target with legible trajectories, though achieving high overall legibility and balancing efficiency remain open challenges for future work.

Abstract

With the increasing presence of social robots in various environments and applications, there is an increasing need for these robots to exhibit socially-compliant behaviors. Legible motion, characterized by the ability of a robot to clearly and quickly convey intentions and goals to the individuals in its vicinity, through its motion, holds significant importance in this context. This will improve the overall user experience and acceptance of robots in human environments. In this paper, we introduce a novel approach to incorporate legibility into local motion planning for mobile robots. This can enable robots to generate legible motions in real-time and dynamic environments. To demonstrate the effectiveness of our proposed methodology, we also provide a robotic stack designed for deploying legibility-aware motion planning in a social robot, by integrating perception and localization components.

Figures (11)

• Figure 1: The robot is tasked with delivering an item to the target person $G^*$, while other people are also present in the scene $\{G_2\}$. A standard (non-legible) local planner would generate the light-green path $\tilde{\xi}^{G^*}$, as it minimizes the planning cost. The proposed legibility-aware planner, however, incorporates a similarity cost term that pushes the planner to generate a path with lower similarity to the unintended path $\tilde{\xi}^{G_2}$ (in red), and it generates the dark green path at the top $\xi^{G^*}$, which decreases the confusion for the observer, and improves the legibility of the robot's motion.
• Figure 2: Generating legible paths using the proposed legibility-aware motion planning algorithm. The sequence of images illustrates the planner's progression towards the goal $G_1$ at time steps 0, 7.2$s$, and 13.2$s$, respectively. Red and green arrows show the pre-computed local path for goals $G_1$ and $G_2$, respectively.
• Figure 3: Comparing the plan generated by the standard planner with cost function $C_{\text{Task}}$ (left) against the output of the legibility-aware motion planning algorithm (right).
• Figure 4: The effect of the field of view cost function $C_{\text{FOV}}$ on the motion plan generated by the legibility-aware planner. The triangles in cyan represent the field of view of the observer, while assuming their depth is infinite.
• Figure 5: The diagram of the robot stack proposed for Legibility-Aware navigation of mobile robots. Nodes and topics are shown by ellipses and rectangles, respectively, and input devices are shown by circles.