Relative State Estimation using Event-Based Propeller Sensing
Ravi Kumar Thakur*, Luis Granados Segura, Jan Klivan, Radim Špetlík Tobiáš Vinklárek, Matouš Vrba, Martin Saska · 2026
Read Relative State Estimation using Event-Based Propeller Sensing by Ravi Kumar Thakur*, Luis Granados Segura, Jan Klivan, Radim Špetlík Tobiáš Vinklárek, Matouš Vrba, Martin Saska (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-18289.txt.
Introduction
Thakur and six colleagues estimate the relative states of drone swarms from event-camera views of spinning propellers. Frame cameras are slow and fail in harsh light, while event cameras offer low latency and microsecond timing. The system tracks propellers in the event stream, turns per-propeller frequencies into thrust inputs, and fits ellipses to recover tilt. On five real outdoor flights, propeller frequency error stays under 3 percent, enabling decentralized localization for multi-robot teams.
Catalog record
| CREATOR | Ravi Kumar Thakur*, Luis Granados Segura, Jan Klivan, Radim Špetlík Tobiáš Vinklárek, Matouš Vrba, Martin Saska |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35232 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.18289
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Relative State Estimation using Event-Based Propeller Sensing Authors are with the Department of Cybernetics, Faculty of Electrical Engineering, Czech Technical University in Prague, Karlovo Namesti 13, 121 35 Prague 2, Czechia. *Corresponding author: [email protected] This work was funded by the Czech Science Foundation (GAČR) under research project no. 26-22419S Ravi Kumar Thakur*, Luis Granados Segura, Jan Klivan, Radim Špetlík Tobiáš Vinklárek, Matouš Vrba, Martin Saska Abstract Autonomous swarms of multi- UAV ( UAV ) system requires an accurate and fast relative state estimation. Although monocular frame-based camera methods perform well in ideal conditions, they are slow, suffer scale ambiguity, and often struggle in visually challenging conditions. The advent of event cameras addresses these challenging tasks by providing low latency, high dynamic range, and microsecond-level temporal resolution. This paper proposes a framework for relative state estimation for quadrotors using event-based propeller sensing. The propellers in the event stream are tracked by detection to extract the region-of-interests. The event streams in these regions are processed in temporal chunks to estimate per-propeller frequencies. These frequency measurements drive a kinematic state estimation module as a thrust input, while camera-derived position measurements provide the update step. Additionally, we use geometric primitives derived from event streams to estimate the orientation of the quadrotor by fitting an ellipse over a propeller and backprojecting it to recover body-frame tilt-axis. The existing event-based approaches for quadrotor state estimation use the propeller frequency in simulated flight sequences. Our approach estimates the propeller frequency under 3 % 3\% error on a test dataset of five real-world outdoor flight sequences, providing a method for decentralized relative localization for multi-robot systems using event cameras. I INTRODUCTION Autonomous formation flight of a multi-robot aerial swarms requires that robots estimate the states of others accurately. In an aerial swarm, the robots can interact with each other to pursue a common goal. Accurate estimation of linear and rotational kinematic states is necessary to enable multi-robot coordination by performing collision avoidance, infer intent, predicting trajectories and maintaining relative poses with respect to each other. This is commonly achieved by explicit communication between them [ 6 ] or by using relative localization through various sensing modalities [ 26 , 24 , 7 ] . Most formation flights involving quadrotors rely on robust communication systems to exchange state information. However, performance can be degraded in the presence of communication latency or packet loss, and in some cases, the stability of formation may be compromised. To solve this problem, we propose event-based propeller sensing to estimate the relative states of quadrotors, thus enabling decentralized swarm flight in environmentally challenging conditions Fig. 1: Visuals of a quadrotor in-flight as seen in pseudo-frame representation of streams accumulated from the event camera. For an autonomous swarm flight to be resilient it needs the quadrotors to interact locally with limited communication, typically based on vision-based perception. However, vision-based methods rely on a frame-based RGB camera which records the light intensity per pixel for every frame. The frame-based camera has low temporal resolution, high latency, and low dynamic range in comparison to event cameras. These limitations make it a less suitable choice for vision-based state estimation where the environment introduces motion blur, extreme illumination changes, and background noise due to clutter. Thus, frame-based cameras introduce limitations for multi-robot formation flight of autonomous aerial swarms in a perceptually challenging environment. To address the limitations of frame-based cameras in multi-robot formation flight of aerial swarms, our work relies on event cameras. These sensors are low-powered and operate at high frequency [ 9 ] . The sensor area is activated only when there is a change in the visual appearance of the scene. The event streams are generated by any motion in the sensor area. Due to these advantages, event cameras are considered for autonomous navigation tasks where there are dynamic obstacles and the quadrotors are moving at high speed [ 25 , 19 ] . Such an autonomous swarm quadrotor system has many applications. It can provide support for humanitarian assistance and disaster relief missions, survey a large area, and gather spatial intelligence. It can enable planetary exploration with less dependence on the orbiter spacecraft. This system is robust against individual failure and provides efficient spatial coverage for exploration. It can reduce dependency on the orbiter spacecraft for navigation. The distributed network of flying robots can act as a synthetic aperture or form a sensor network [ 20 ] . Heterogeneous robot teams are proposed for the exploration of canyon-like structures where communication is challenging [ 23 , 17 ] . In this paper, we propose the use of a neuromorphic camera as the primary sensor to perform relative state estimation. Our approach utilizes the event streams from the camera to detect the quadrotor, localize it, and estimate the frequency of its propellers. The frequency of the propeller is used by the state estimation pipeline to get first-order kinematics of the quadrotor. We also make use of the shape of the propeller disk detected in the event streams to estimate the orientation of the quadrotor using P1E ( P1E ) [ 10 ] . II RELATED WORK II-A Frequency Detection of Periodic Phenomena The high dynamic range, low latency and microsecond resolution make event cameras suitable for studying periodic events. They provide a non-invasive way to perform measurements. A kernel-based approach was proposed to estimate angular speeds by manually selecting regions in the event data of a rotating object [ 3 ] . To estimate frequency of any mechanical periodic phenomena like rotation, flicker, vibration etc. [ 11 ] proposed generation of correlation peaks with automatic template selection. In [ 27 ] a combined heatmap and clustering based approach is proposed for propeller detection and subsequently frequency estimation using ICP ( ICP ) registration. [ 16 ] proposed the use of digital infinite impulse response filter for brightness reconstruction and created a per-pixel frequency map by detecting zero-level crossing. A method for creating frequency maps is proposed by applying the FFT ( FFT ) to the event stream of each pixel [ 1 ] . These approaches assume a completely periodic motion without any drift or ego motion, thus limiting their application in formation flight. II-B State Estimation using Event Camera The estimation of state of UAV is of large interest in swarm robotics. Using an RGB camera for estimation of the relative pose of the camera is a common approach. However, a similar approach cannot be applied to the event camera. The potential use of propeller sensing to estimate rotational attitude parameters was proposed in [ 22 ] . In this approach, the event streams are treated as the Poisson process to distinguish events generated by a propeller and the background. The pose is estimated by fitting an ellipse over the aggregated propeller events. HelixTrack extends this work by estimating the pose of the roating propeller using a batched Gauss-Newton optimization. They test their algorithm on propeller dataset with different levels of ego-motion [ 21 ] . Eventpro [ 5 ] proposes the estimation of the propeller frequency using hierarchical event processing and known propeller geometry. The flight commands are then predicted using a trained SVM ( SVM ) classifier. In [ 13 ] , an