CHORAL: Traversal-Aware Planning for Safe and Efficient Heterogeneous Multi-Robot Routing

TL;DR

This paper proposes an integrated semantic-aware framework for coordinating heterogeneous robots, CHORAL, and releases it as open source to support reproducibility and deployment of diverse robot teams.

Abstract

Monitoring large, unknown, and complex environments with autonomous robots poses significant navigation challenges, where deploying teams of heterogeneous robots with complementary capabilities can substantially improve both mission performance and feasibility. However, effectively modeling how different robotic platforms interact with the environment requires rich, semantic scene understanding. Despite this, existing approaches often assume homogeneous robot teams or focus on discrete task compatibility rather than continuous routing. Consequently, scene understanding is not fully integrated into routing decisions, limiting their ability to adapt to the environment and to leverage each robot's strengths. In this paper, we propose an integrated semantic-aware framework for coordinating heterogeneous robots. Starting from a reconnaissance flight, we build a metric-semantic map using open-vocabulary vision models and use it to identify regions requiring closer inspection and capability-aware paths for each platform to reach them. These are then incorporated into a heterogeneous vehicle routing formulation that jointly assigns inspection tasks and computes robot trajectories. Experiments in simulation and in a real inspection mission with three robotic platforms demonstrate the effectiveness of our approach in planning safer and more efficient routes by explicitly accounting for each platform's navigation capabilities. We release our framework, CHORAL, as open source to support reproducibility and deployment of diverse robot teams.

Figures (9)

• Figure 1: Our framework for heterogeneous multi-robot inspection in a real-world deployment. A metric-semantic map of the environment is built from a reconnaissance flight, enabling the automatic identification of inspection targets and the computation of platform-specific, traversal-aware routes. Ground robots avoid regions with poor traversability with pebbles or cables, while aerial robots plan safer trajectories that maintain clearance from obstacles.
• Figure 2: Overview of our framework for heterogeneous multi-robot routing. (a) An inspection mission is defined in an unknown environment based on user-specified task classes. (b) A surveying aerial platform constructs a metric-semantic map using open-vocabulary vision models. (c) The map is processed to identify inspection targets, compute feasible connections between tasks, and extract platform-specific traversal costs. (d) A heterogeneous vehicle routing problem is formulated and solved to generate safe and efficient routes tailored to the capabilities of each robot. (e) The resulting plans are executed using standardized navigation stacks.
• Figure 3: Open vocabulary map built with our mapping module, showing the flexibility in specifying semantically rich classes.
• Figure 4: PRM construction. From left to right: the tasks are shown in blue and obstacles in black, the covisible tasks are connected, disjoint sets are detected and merged using RRT*.
• Figure 5: Diagram of the ROS 2 system implementation provided, indicating the components in our framework. Yellow means the component was fully developed for this article. Gray means the component was adapted and integrated into our system.