Abstract
Lightweight aerial swarms have potential applications in scenarios where larger drones fail to operate efficiently. The primary foundation for lightweight aerial swarms is efficient relative localization, which enables cooperation and collision avoidance. Computing the real-time position is challenging due to extreme resource constraints. This paper presents an autonomous relative localization technique for lightweight aerial swarms without infrastructure by fusing ultra-wideband wireless distance measurements and the shared state information (e.g., velocity, yaw rate, height) from neighbors. This is the first fully autonomous, tiny, fast, and accurate relative localization scheme implemented on a team of 13 lightweight (33 grams) and resource-constrained (168MHz MCU with 192 KB memory) aerial vehicles. The proposed resource-constrained swarm ranging protocol is scalable, and a surprising theoretical result is discovered: the unobservability poses no issues because the state drift leads to control actions that make the state observable again. By experiment, less than 0.2m position error is achieved at the frequency of 16Hz for as many as 13 drones.
What Problem Does This Paper Address?
For a swarm of drones to fly in formation or avoid collisions, every drone needs to know where its neighbors are. While large drones can carry heavy Lidars or powerful computers for visual processing, "micro-flyers" like the Crazyflie 2.1 are severely limited. They weigh about as much as a lightbulb and run on microcontrollers with less memory than a 1990s calculator.
Existing solutions usually fall into two traps: they either require pre-installed "anchors" (beacons) in the room, or they use ranging protocols that don't scale—as you add more drones, the communication "chatter" becomes so slow that the swarm becomes unstable. Furthermore, there is a mathematical "unobservability" problem: in certain flight patterns, a drone’s sensors literally cannot distinguish between different relative positions, leading to dangerous estimation drifts.
Methods
The authors tackle these constraints through a three-pronged approach, integrating onboard sensing with a novel communication strategy.
- Scalable UWB Ranging Protocol: Instead of traditional one-to-one "polling" (which is slow), they designed a broadcast-based protocol. Each drone periodically broadcasts a single message containing its own state and the timestamps of messages it received from others. This allows every drone in the vicinity to calculate distances to all its neighbors concurrently.
- Tiny EKF (Extended Kalman Filter): They implemented a highly optimized EKF that fuses UWB distance measurements with shared state data (velocity, yaw rate, and height) sent by neighboring drones. This runs entirely onboard the drone's 168MHz processor.
- Stochastic Initialization: Since drones start at unknown positions, the system uses a brief "random flight" phase. These random movements provide the necessary geometric variety for the EKF to "lock on" to the correct relative positions of neighbors.
Key Insights and Results
Self-Regulated Convergence
The researchers discovered a "self-regulating" property of the system. Mathematically, if a drone flies in a perfect, steady formation, part of its state becomes "unobservable," meaning the localization filter should technically fail. However, if the estimate starts to drift, the drone's control law (which relies on that estimate) inadvertently causes the drone to move in a way that makes the state observable again. The system essentially "fixes" its own lack of data through movement.
Performance Metrics
- High Precision: The system achieved a position error of less than 0.2 meters (averaging around 0.13m in real-world tests).
- High Frequency: The swarm maintained a 16Hz update rate for 13 drones, which is fast enough for stable, agile flight.
- Scalability: The proposed protocol is roughly 5 times more efficient than standard token-ring algorithms when 9 or more drones are participating.
Conclusion
This work demonstrates that high-precision relative localization is possible on extremely resource-constrained hardware without external infrastructure. By leveraging a scalable broadcast protocol and the inherent coupling between control and observability, the researchers enabled a swarm of 13 tiny drones to perform complex autonomous tasks. The code is open-sourced, providing a foundation for future research in resource-restricted robotics.
Summary updated