Researchers unveil bearing-based circumnavigation with collision avoidance for drone swarms
New algorithm lets drone swarms orbit a target using only relative bearings, no GPS needed.
Researchers Kushal Pratap Singh, Twinkle Tripathy, and Anoop Jain have developed a distributed guidance law that enables a heterogeneous team of agents to circumnavigate a stationary target while avoiding collisions — even when only a subset of agents (leaders) know the target's actual location. The key innovation is that each agent is modeled as a disk (not a point mass), accounting for physical dimensions, and the control law relies solely on relative bearing measurements from designated nearest neighbors in a time-varying interaction graph. By controlling only angular speeds, the system achieves stable circumnavigation without requiring distance or global position information.
The proposed method leverages a logarithmic Barrier Lyapunov Function (BLF) to guarantee forward invariance of the collision-free set — meaning agents never collide if they start in admissible positions. The approach works for both static and time-varying communication topologies, making it robust to network changes. Simulations on 13 pages with 27 figures show that all followers converge to circular orbits around the same target as the leader at the end of their directed path. This work offers a practical, scalable solution for decentralized drone swarms or robotic teams operating in GPS-denied environments with limited sensing.
- Agents are modeled as disks (not point masses) to account for physical dimensions and ensure collision avoidance.
- Only angular speeds are controlled; no distance or global coordinates needed — just bearing from nearest neighbors.
- Collision avoidance is mathematically guaranteed via a logarithmic Barrier Lyapunov Function, ensuring forward invariance of safe configurations.
Why It Matters
Enables scalable, decentralized drone swarms for surveillance or target tracking without GPS or centralized coordination.