Heterogeneous robot teams distribute complementary capabilities across specialized agents, but their physical roles and capacities typically remain fixed throughout a mission. We present HARP, a Heterogeneous Aerial Robotic modules Platform in which independently deployable aerial robots physically reconfigure to compose their capabilities for field operations. HARP comprises sensor-equipped scouts, flydrive rover modules, and task-specific payload modules. Scouts map the environment and inform an energy-aware planner that jointly selects routes and air-ground mobility modes. Rover and payload modules fly independently across terrain that constrains ground travel, then autonomously assemble into a cooperative ground vehicle for energy-efficient payload transport. Motivated by environmental sampling in remote and difficult-to-traverse regions, we evaluate HARP through field experiments spanning sensing, planning, reconfiguration, airground mobility, payload transport, and task execution. We further conduct module-level deployment tests on the Greenland Ice Sheet toward future autonomous missions. HARP demonstrates how heterogeneous robot teams can adapt not only their actions, but also how their physical capabilities are composed during a mission.
Figures & tables
Fig. 1: HARP modules undergoing unit field tests at the margin of the Greenland Ice Sheet near Kangerlussuaq Greenland. The system has three types of aerial modules: scouts for terrain mapping and path planning, rovers for ground traversal and payload transport, and payloads for mission-specific tasks.
Fig. 2: Overview of the HARP system architecture and modular configurations.
Fig. 3: Terrain mapping and joint air-ground planning across three representative environments. Top: scout-generated LiDAR maps colored by elevation. Bottom: cost-to-go V (yellow low, blue–purple high) over the ground and aerial mobility layers with blocked states shown in gray. Red, cyan, and yellow denote routes and initial positions 1–3, and the white diamond is the shared ground goal. Vertical segments indicate mode transitions.
Fig. 4: Physical benefits of reconfiguration. (a) Model-estimated energy drawn from the payload or integrated task-bearing platform before task execution, averaged across three initial conditions. The ground-only baseline has no feasible route under the terrain and collision constraints. (b) Measured tractive force by a single rover and by the four-rover assembled configuration.
Fig. 5: Autonomous vertical-docking performance over 11 repeated trials. Curves show payload trajectories during terminal descent toward the center of the four-rover formation. Circles denote successful capture and locking, while crosses denote unsuccessful trials.
Fig. 6: End-to-end outdoor field demonstration. (a) Executed trajectory over the reconstructed terrain map, with assembled ground and independent aerial segments. (b) Subsurface sample recovered by the drilling payload. (c) Representative sequence showing ground traversal, disassembly, formation flight, vertical docking, and reassembly.
Department of Electrical and Computer Engineering, National University of Singapore, Singapore 117583 · Department of Mechanical Engineering, National University of Singapore, Singapore 117583