Aerial Swarm Research Articles

Unsupervised Contrastive Learning for Automatic Grouping of Aerial Swarms

Unsupervised Contrastive Learning for Automatic Grouping of Aerial Swarms

A collaborative navigation integrity augmentation method with optimal dynamic partner selection

The collaboration among swarmed aircraft can provide additional observation to improve the integrity of its onboard navigation systems. But due to the limitation on relative communicating and measuring burdens, adopting all the observations between aircraft in the swarm is inefficient. The geometry of the collaborated partners is a key factor that influences the effectiveness of the navigation integrity augmentation, which needs to be optimized in collaborative integrity augmented navigation. In this paper, the integrity augmented navigation method for aerial swarm based on collaborative partner optimization is proposed. The geometry constructed by the GNSS satellites and the cooperative partners in the aerial swarm are analyzed dynamically, and the augmented integrity protection levels with different collaborative relationships are predicted to distinguish the partner essential for navigation integrity augmentation. Then the collaborative partner makes a key contribution to integrity augmentation adopted in collaborative navigation integrity monitoring, improving the efficiency of the collaborative navigation. The simulation results indicate the effectiveness of the proposed cooperative partnership optimization strategy, as well as the superiority of the proposed method compared with the traditional independent integrity framework in improving the integrity protection level and fault detection capacity.

Collaborative Localization Method Based on Hybrid Network for Aerial Swarm

In light of the satellite rejection environment and how aircraft can obtain high-precision positioning, this paper proposes a collaborative correction algorithm for aircraft based on the rank-defect network. Aiming at the problem of insufficient anchor points, which result in insufficient observations and the divergence of aircraft inertial navigation errors, this algorithm can effectively improve the navigation performance of cluster aircraft. On the basis of the observation information provided by the anchor aircraft, the observation information between aircraft is fully utilized to improve the observability of the aircraft cluster positioning method. At the same time, the pseudo-observation equation of heterogeneous aircraft cluster positioning is introduced, and the divergence of inertial navigation positioning errors caused by insufficient observations is suppressed by the pseudo-observation solution. On the basis of introducing the pseudo-observation equation, the inertial navigation error is solved and corrected by the Newton iterative method and the divergence of the inertial navigation position error is restrained. Compared with an aircraft cluster positioning method that does not use the inertial navigation error co-correction based on the pseudo-observation solution, this paper can achieve better overall cluster positioning accuracy when the available observations are insufficient, which is suitable for practical applications.

Using modified trapping regimes to understand the behavioral and spatial ecology of Philornis downsi (Diptera: Muscidae).

The avian vampire fly Philornis downsi (Dodge & Aitken) (Diptera: Muscidae) is native to continental South America and the Caribbean, but invasive in the Galapagos Archipelago. The larvae of P. downsi feed on the blood and tissues of the nestlings of 75% of the small land bird species that are endemic or native to Galapagos, causing high in-nest mortality and severe population declines in some species. Efficient trapping techniques are vital to safeguarding these birds in the short term as well as for monitoring fly populations, but basic information about the ecology of the fly is still needed to help develop a species-appropriate trapping method. In this study, we used a novel trapping regime with a vertical distribution to make inferences about P. downsi's behavioral and spatial ecology and to optimize trap catch. Our results showed that male and female P. downsi were trapped in greater numbers below the canopy (3.1-7.5 m), lower down than other commonly caught insect species (5.1-11.5 m). Notably, the effect of trap height remained consistent across seasons and different weather conditions. These findings suggest that P. downsi tend to move at heights where their hosts nest (at or below the canopy) and do not spend time above the canopy. This also makes it unlikely that strategies such as hill-topping or aerial swarming are being used to locate mates. As such, trapping and control efforts should be focused below the canopy in forests with similar canopy heights to effectively capture P. downsi and reduce bycatch of other insects.

$D^{2}$SLAM: Decentralized and Distributed Collaborative Visual-Inertial SLAM System for Aerial Swarm

$D^{2}$SLAM: Decentralized and Distributed Collaborative Visual-Inertial SLAM System for Aerial Swarm

Distributed TDMA Scheduling for Autonomous Aerial Swarms: A Self-Organizing Approach

Distributed TDMA Scheduling for Autonomous Aerial Swarms: A Self-Organizing Approach

High-Speed Motion Planning for Aerial Swarms in Unknown and Cluttered Environments

High-Speed Motion Planning for Aerial Swarms in Unknown and Cluttered Environments

Application of Deep Reinforcement Learning to UAV Swarming for Ground Surveillance.

This paper summarizes in depth the state of the art of aerial swarms, covering both classical and new reinforcement-learning-based approaches for their management. Then, it proposes a hybrid AI system, integrating deep reinforcement learning in a multi-agent centralized swarm architecture. The proposed system is tailored to perform surveillance of a specific area, searching and tracking ground targets, for security and law enforcement applications. The swarm is governed by a central swarm controller responsible for distributing different search and tracking tasks among the cooperating UAVs. Each UAV agent is then controlled by a collection of cooperative sub-agents, whose behaviors have been trained using different deep reinforcement learning models, tailored for the different task types proposed by the swarm controller. More specifically, proximal policy optimization (PPO) algorithms were used to train the agents' behavior. In addition, several metrics to assess the performance of the swarm in this application were defined. The results obtained through simulation show that our system searches the operation area effectively, acquires the targets in a reasonable time, and is capable of tracking them continuously and consistently.

Edge computing powers aerial swarms in sensing, communication, and planning

Edge computing powers aerial swarms in sensing, communication, and planning

Efficiently Range-Coupled Position Estimation for an Aerial Swarm With the Confidence Evaluation of Onboard Sensing

This work proposes an efficient range-coupled localization approach for aerial swarms considering the variance of onboard sensors in cluttered environments. First, a reciprocal bond estimation algorithm is proposed to estimate the position; it constructs an optimization framework based on the range-coupled observability analysis. Subsequently, considering the variable performance characteristics of onboard sensors in cluttered environments, a moving confidence evaluation algorithm is proposed to improve the resilience of the position estimation system by assessing the concurrent reliability of all sensors. Mathematically, the aforementioned two algorithms are integrally formulated using a nonlinear least-squares equation. Notably, solving this nonlinear problem is typically a low-efficiency process that deteriorates with the scale of the swarm. To overcome this issue, a gradient-aware Levenberg–Marquardt algorithm is herein proposed to enhance the computational efficiency of this system to solve this nonlinear least-squares problem. Finally, swirling experiments involving four drones are carried out to verify the performance of the proposed methods. The results demonstrate that the time cost of optimization in each servo period is only approximately 7.15 ms, and the computational efficiency exhibits at least a 2.78 times improvement over the existing methods. Meanwhile, even in a smoky environment, the estimation precision can be as great as 8 cm, which is comparable to results obtained using state-of-the-art methods.

Technological Perspectives of Countering UAV Swarms

Conventional AD systems have been found less effective for countering UAVs and loitering munitions. Thishas necessitated the development of counter-UAV systems with different functionalities. A cluster of armed UAVsas swarm formations has further rendered the conventional AD systems far from effective, emphasizing the need to consider countering swarms as the most crucial element in new-generation aerial threat mitigation strategies. In this paper, the capabilities of UAV swarms and vital military assets exposed to such attacks are identified. To protect the vital assets from aerial swarm threats, ideal system characteristics of a counter-UAV (C-UAV) swarm system to overcome the challenges are discussed. Currently available acquisition & engagement technology is analyzed and the application of these systems to counter swarm applications is brought out. New requirements are discussed and a conceptual design of a layered system is derived which can handle a large spectrum of aerial threats including a swarm of UAVs. This system is expected to have a higher rate of engagement and can be designed with low-cost network-integrated systems.

Aerial robot swarms: a review

<span lang="EN-US">The purpose of this review is to highlight the current research on aerial robot swarms and their applications. It focuses on the system architecture and follows the current trend in aerial robotics promoting research in this field along with its impact on society. Further, it explores the dynamics as well as the flying mechanisms of a drone and sheds light on the different algorithms being used to control aerial swarms. Due to a lot of research going on in this field, we also discuss the different trends that are active and of keen interest to the researchers, including the swarm pattern formation behavior.</span>

Omni-Swarm: A Decentralized Omnidirectional Visual–Inertial–UWB State Estimation System for Aerial Swarms

Decentralized state estimation is one of the most fundamental components of autonomous aerial swarm systems in GPS-denied areas yet it still remains a highly challenging research topic. Omni-swarm, a decentralized omnidirectional visual-inertial-UWB state estimation system for aerial swarms, is proposed in this paper to address this research niche. To solve the issues of observability, complicated initialization, insufficient accuracy, and lack of global consistency, we introduce an omnidirectional perception front-end in Omni-swarm. It consists of stereo wide-FoV cameras and ultra-wideband sensors, visual-inertial odometry, multi-drone map-based localization, and visual drone tracking algorithms. The measurements from the front-end are fused with graph-based optimization in the back-end. The proposed method achieves centimeter-level relative state estimation accuracy while guaranteeing global consistency in the aerial swarm, as evidenced by the experimental results. Moreover, supported by Omni-swarm, inter-drone collision avoidance can be accomplished without any external devices, demonstrating the potential of Omni-swarm as the foundation of autonomous aerial swarms.

Adaptive arbitration of aerial swarm interactions through a Gaussian kernel for coherent group motion.

Swarm behaviors offer scalability and robustness to failure through a decentralized and distributed design. When designing coherent group motion as in swarm flocking, virtual potential functions are a widely used mechanism to ensure the aforementioned properties. However, arbitrating through different virtual potential sources in real-time has proven to be difficult. Such arbitration is often affected by fine tuning of the control parameters used to select among the different sources and by manually set cut-offs used to achieve a balance between stability and velocity. A reliance on parameter tuning makes these methods not ideal for field operations of aerial drones which are characterized by fast non-linear dynamics hindering the stability of potential functions designed for slower dynamics. A situation that is further exacerbated by parameters that are fine-tuned in the lab is often not appropriate to achieve satisfying performances on the field. In this work, we investigate the problem of dynamic tuning of local interactions in a swarm of aerial vehicles with the objective of tackling the stability-velocity trade-off. We let the focal agent autonomously and adaptively decide which source of local information to prioritize and at which degree-for example, which neighbor interaction or goal direction. The main novelty of the proposed method lies in a Gaussian kernel used to regulate the importance of each element in the swarm scheme. Each agent in the swarm relies on such a mechanism at every algorithmic iteration and uses it to tune the final output velocities. We show that the presented approach can achieve cohesive flocking while at the same time navigating through a set of way-points at speed. In addition, the proposed method allows to achieve other desired field properties such as automatic group splitting and joining over long distances. The aforementioned properties have been empirically proven by an extensive set of simulated and field experiments, in communication-full and communication-less scenarios. Moreover, the presented approach has been proven to be robust to failures, intermittent communication, and noisy perceptions.

A Mean-Field Game Control for Large-Scale Swarm Formation Flight in Dense Environments.

As an important part of cyberphysical systems (CPSs), multiple aerial drone systems are widely used in various scenarios, and research scenarios are becoming increasingly complex. However, planning strategies for the formation flying of aerial swarms in dense environments typically lack the capability of large-scale breakthrough because the amount of communication and computation required for swarm control grows exponentially with scale. To address this deficiency, we present a mean-field game (MFG) control-based method that ensures collision-free trajectory generation for the formation flight of a large-scale swarm. In this paper, two types of differentiable mean-field terms are proposed to quantify the overall similarity distance between large-scale 3-D formations and the potential energy value of dense 3-D obstacles, respectively. We then formulate these two terms into a mean-field game control framework, which minimizes energy cost, formation similarity error, and collision penalty under the dynamical constraints, so as to achieve spatiotemporal planning for the desired trajectory. The classical task of a distributed large-scale aerial swarm system is simulated by numerical examples, and the feasibility and effectiveness of our method are verified and analyzed. The comparison with baseline methods shows the advanced nature of our method.

Fast Joint Multi-Robot Trajectory Optimization by GPU Accelerated Batch Solution of Distributed Sub-Problems.

We present a joint multi-robot trajectory optimizer that can compute trajectories for tens of robots in aerial swarms within a small fraction of a second. The computational efficiency of our approach is built on breaking the per-iteration computation of the joint optimization into smaller, decoupled sub-problems and solving them in parallel through a custom batch optimizer. We show that each of the sub-problems can be reformulated to have a special Quadratic Programming structure, wherein the matrices are shared across all the problems and only the associated vector varies. As result, the batch solution update rule reduces to computing just large matrix vector products which can be trivially accelerated using GPUs. We validate our optimizer’s performance in difficult benchmark scenarios and compare it against existing state-of-the-art approaches. We demonstrate remarkable improvements in computation time its scaling with respect to the number of robots. Moreover, we also perform better in trajectory quality as measured by smoothness and arc-length metrics.

Agile Formation Control of Drone Flocking Enhanced With Active Vision-Based Relative Localization

The vision-based relative localization can provide effective feedback for the cooperation of aerial swarm and has been widely investigated in previous works. However, the limited field of view (FOV) inherently restricts its performance. To cope with this issue, we propose a novel distributed active vision-based relative localization framework and apply it to formation control in aerial swarms. Inspired by bird flocks in nature, we realize omnidirectional visual detection with a novel active vision structure and devise graph-based attention planning (GAP) to obtain the optimal observation target assignment for the swarm’s active vision. Then active visual detection results are fused with onboard measurements from Ultra-WideBand (UWB) and visual-inertial odometry (VIO) to obtain real-time relative positions, which further improve the formation control performance of the swarm. Simulations and experiments demonstrate that the proposed active vision system outperforms the fixed vision system in terms of estimation and formation accuracy.

Fault-tolerant cooperative navigation of networked UAV swarms for forest fire monitoring

Coordination of unmanned aerial vehicle (UAV) swarms has received significant attention due to its wide practical applications including search and rescue, cooperative exploration and target surveillance. Motivated by the flexibility of the UAVs and the recent advancement of graph-based cooperative control strategies, this paper aims to develop a fault-tolerant cooperation framework for networked UAVs with applications to forest fire monitoring. Firstly, a cooperative navigation strategy based on network graph theory is proposed to coordinate all the connected UAVs in a swarm in the presence of unknown disturbances. The stability of the aerial swarm system is guaranteed using the Lyapunov approach. In case of damage to the actuators of some of the UAVs during the mission, a decentralized task reassignment algorithm is then applied, which makes the UAV swarm more robust to uncertainties. Finally, a novel geometry-based collision avoidance approach using onboard sensory information is proposed to avoid potential collisions during the mission. The effectiveness and feasibility of the proposed framework are verified initially by simulations and then using real-world flight tests in outdoor environments.

Decentralised aerial swarm for adaptive and energy efficient transport of unknown loads

• We present a behaviour based subsumption architecture that enables a swarm’s topology to adapt to a load's mass distribution. • The decentralized self-organized swarm can perform a cooperative load transportation task without any explicit communication between members. • Our architecture has the potential to be scalable to a large number of agents. • This is significant in aerial swarm cooperative transport systems because it leads to even energy distribution among swarm members when compared to a uniform swarm topology. • It also fosters robustness of the load-suspended aerial robot system. Cooperative transport by a swarm of Quadcopters offers more flexibility and performance when carrying loads that are complex in structural profile and mass. Ensuring that team members of the swarm are optimally placed on these loads as well as able to resist disturbances from the environment during transport are current research challenges. In this paper, we present a decentralized behaviour based subsumption architecture for enabling a swarm of Quadcopters to explore an unfamiliar area, find a load and transport it to a target location cooperatively. In the architecture, three behaviours were used: an obstacle avoidance behaviour to avoid collisions with objects in the environment, a flocking behaviour to ensure swarm structure and a bacterium behaviour for exploration of the environment and to adapt to the mass profile of various detected loads. By adapting to the mass profile of a detected load, we show that our architecture ensures even energy distribution among Quadcopters while achieving robustness to disturbances from the environment. Our results show that a mass adapting swarm is able to conserve energy during payload transportation when compared to a swarm that does not adapt to a load’s profile. Furthermore, we do not use explicit communication between team members but instead rely on data from visual sensors attached to the Quadcopters. We experiment with simulations in a physics informed robot simulator called CoppeliaSim and demonstrate the effectiveness of our architecture when utilized for cooperative transport of irregular loads.

Neural Tree Expansion for Multi-Robot Planning in Non-Cooperative Environments

We present a self-improving, Neural Tree Expansion (NTE) method for multi-robot online planning in non-cooperative environments, where each robot attempts to maximize its cumulative reward while interacting with other self-interested robots. Our algorithm adapts the centralized, perfect information, discrete-action space method from AlphaZero to a decentralized, partial information, continuous action space setting for multi-robot applications. Our method has three interacting components: (i) a centralized, perfect-information “expert” Monte Carlo Tree Search (MCTS) with large computation resources that provides expert demonstrations, (ii) a decentralized, partial-information “learner” MCTS with small computation resources that runs in real-time and provides self-play examples, and (iii) policy & value neural networks that are trained with the expert demonstrations and bias both the expert and the learner tree growth. Our numerical experiments demonstrate Neural Tree Expansion's computational advantage by finding better solutions than a MCTS with 20 times more resources. The resulting policies are dynamically sophisticated, demonstrate coordination between robots, and play the Reach-Target-Avoid differential game significantly better than the state-of-the-art control-theoretic baseline for multi-robot, double-integrator systems. Our hardware experiments on an aerial swarm demonstrate the computational advantage of Neural Tree Expansion, enabling online planning at 20 Hz with effective policies in complex scenarios.