Quadrotor Swarm Research Articles

Learning Scalable Decentralized Controllers for Heterogeneous Robot Swarms with Graph Neural Networks

Abstract Distributed multi-agent systems are becoming increasingly crucial for diverse applications in robotics because of their capacity for scalability, efficiency, robustness, resilience, and the ability to accomplish complex tasks. Controlling these large-scale swarms by relying on local information is very challenging. Although centralized methods are generally efficient or optimal, they face the issue of scalability and are often impractical. Given the challenge of finding an efficient decentralized controller that uses only local information to accomplish a global task, we propose a learning-based approach to decentralized control using supervised learning. Our approach entails training controllers to imitate a centralized controller's behavior but uses only local information to make decisions. The controller is parameterized by aggregation graph neural networks (GNNs) that integrate information from remote neighbors. The problems of segregation and aggregation of a swarm of heterogeneous agents are explored in 2D and 3D point mass systems as two use cases to illustrate the effectiveness of the proposed framework. The decentralized controller is trained using data from a centralized (expert) controller derived from the concept of artificial differential potential. Our learned models successfully transfer to actual robot dynamics in physics-based Turtlebot3 robot swarms in Gazebo/ROS2 simulations and hardware implementation and Crazyflie quadrotor swarms in Pybullet simulations. Our experiments show that our controller performs comparably to the centralized controller and demonstrates superior performance compared to a local controller. Additionally, we showed that the controller is scalable by analyzing larger teams and diverse groups up to 100 robots.

Cooperative Safe Trajectory Planning for Quadrotor Swarms.

In this paper, we propose a novel distributed algorithm based on model predictive control and alternating direction multiplier method (DMPC-ADMM) for cooperative trajectory planning of quadrotor swarms. First, a receding horizon trajectory planning optimization problem is constructed, in which the differential flatness property is used to deal with the nonlinear dynamics of quadrotors while we design a relaxed form of the discrete-time control barrier function (DCBF) constraint to balance feasibility and safety. Then, we decompose the original trajectory planning problem by ADMM and solve it in a fully distributed manner with peer-to-peer communication, which induces the quadrotors within the communication range to reach a consensus on their future trajectories to enhance safety. In addition, an event-triggered mechanism is designed to reduce the communication overhead. The simulation results verify that the trajectories generated by our method are real-time, safe, and smooth. A comprehensive comparison with the centralized strategy and several other distributed strategies in terms of real-time, safety, and feasibility verifies that our method is more suitable for the trajectory planning of large-scale quadrotor swarms.

Optimal formation strategy for interconnected quadrotor drones: Theory and experiment

The formation control for a quadrotor swarm with the energy consumption constraint and/or the regulation performance constraint is challenging, where both the guaranteed-performance formation strategy and the guaranteed-cost formation strategy cannot realize the optimal control. Especially, the flying experiment of the optimal time-varying formation is very difficult to realize. A new control protocol with an optimization index is proposed to realize the optimal time-varying formation and the optimization index is integrated as a Kronecker product form containing the Laplacian matrix of the communication topology of a quadrotor swarm. Moreover, an analytic criterion for optimal time-varying formation achievability is proposed, and explicit expressions of the formation center function and the minimum value of the optimization index are presented, respectively. Furthermore, an optimal formation achievability algorithm is proposed and a formation flying experiment is performed, where the outer-loop position control input is converted into the inner-loop attitude control and an Euler-angle loop controller and an angular velocity loop controller are introduced.

DLSC: Distributed Multi-Agent Trajectory Planning in Maze-Like Dynamic Environments Using Linear Safe Corridor

This article presents an online distributed trajectory planning algorithm for a quadrotor swarm in a maze-like dynamic environment. We utilize a dynamic linear safe corridor to construct the feasible collision constraints that can ensure interagent collision avoidance and consider the uncertainty of moving obstacles. We introduce mode-based subgoal planning to resolve deadlock faster in a complex environment using only previously shared information. For dynamic obstacle avoidance, we adopt heuristic methods such as collision alert propagation and escape point planning to deal with the situation where dynamic obstacles approach the agents clustered in a narrow corridor. We prove that the proposed algorithm guarantees the feasibility of the optimization problem for every replanning step. In an obstacle-free space, the proposed method can compute the trajectories for 60 agents on average 7.66 ms per agent with an Intel i7 laptop and shows the perfect success rate. Also, our method shows 64.5 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\%$</tex-math></inline-formula> shorter flight time than buffered Voronoi cell and 34.6 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\%$</tex-math></inline-formula> shorter than with our previous work. We conduct the simulation in a random forest and maze with four dynamic obstacles, and the proposed algorithm shows the highest success rate and shortest flight time compared to state-of-the-art baseline algorithms. In particular, the proposed algorithm shows over 97 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\%$</tex-math></inline-formula> success rate when the velocity of moving obstacles is below the agent's maximum speed. We validate the safety and robustness of the proposed algorithm through a hardware demonstration with ten quadrotors and two pedestrians in a maze-like environment.

Online Distributed Trajectory Planning for Quadrotor Swarm With Feasibility Guarantee Using Linear Safe Corridor

This letter presents a new online multi-agent trajectory planning algorithm that guarantees to generate safe, dynamically feasible trajectories in a cluttered environment. The proposed algorithm utilizes a linear safe corridor (LSC) to formulate the distributed trajectory optimization problem with only feasible constraints, so it does not resort to slack variables or soft constraints to avoid optimization failure. We adopt a priority-based goal planning method to prevent the deadlock without an additional procedure to decide which robot to yield. The proposed algorithm can compute the trajectories for 60 agents on average 15.5 ms per agent with an Intel i7 laptop and shows a similar flight distance and distance compared to the baselines based on soft constraints. We verified that the proposed method can reach the goal without deadlock in both the random forest and the indoor space, and we validated the safety and operability of the proposed algorithm through a real flight test with ten quadrotors in a maze-like environment.

Decentralized aggregation and leader-following control of a swarm of quadcopters with nonlinear under-actuated dynamics

This paper studies the decentralized control of a swarm of quadrotors on the basis of the nonlinear, highly-coupled, and under-actuated dynamic model of quadcopters. The swarm of quadcopters must illustrate the desired swarm behavior as fast as possible while collision avoidance is preserved during the entire evolution process. The interaction relationship among the swarm members is modeled by artificial potential functions and according to the nearest neighbor rule. The sliding mode control technique is employed to control the velocity of quadcopters in the swarm. Finally, two fundamental swarm behaviors, i.e. swarm aggregation and leader-following, are numerically simulated to demonstrate the efficacy of the proposed algorithm.

Multi-objective optimization of a quadrotor flock performing target zone search

Abstract This article presents a performance improvement of a bio-inspired quadrotor swarm for a flocking task with an unknown target zone. Because swarm behavior depends on repulsion, orientation and attraction tendencies between members, these parameters can be set properly for this improvement. A simulator of the swarm in the proposed task is made involving a dynamic modelling of a quadrotor and considering two scenarios, without and with obstacles. Task execution is evaluated by multiple objective functions which must be minimized making use of multi-objective optimization techniques. A comparison between Nondominated Sorting Genetic Algorithm II with Differential Evolution (NSGA-II-DE) and Multi-Objective Particle Swarm Optimization (MOPSO) is made. The results indicate relevance of the proper set of repulsion, orientation and attraction parameters in order to maintain the properties of the swarm like aggregation while executes a navigation-search task when scenario involves a different number of quadrotors or when there are obstacles in the arena.

DCAD: Decentralized Collision Avoidance With Dynamics Constraints for Agile Quadrotor Swarms

We present DCAD, a novel, decentralized collision avoidance algorithm for navigating a swarm of quadrotors in dense environments populated with static and dynamic obstacles. Our algorithm relies on the concept of Optimal Reciprocal Collision Avoidance (ORCA) and utilizes a flatness-based Model Predictive Control (MPC) to generate local collision-free trajectories for each quadrotor. We feedforward linearize the non-linear dynamics of the quadrotor and subsequently use this linearized model in our MPC framework. Our approach tends to compute safe trajectories that avoid quadrotors from entering each other's downwash regions during close proximity maneuvers. In addition, we account for the uncertainty in the position and velocity sensor data using Kalman filter. We evaluate the performance of our algorithm with other state-of-the-art decentralized methods and demonstrate its superior performance in terms of smoothness of generated trajectories and lower probability of collision during high-velocity maneuvers.

LSwarm: Efficient Collision Avoidance for Large Swarms With Coverage Constraints in Complex Urban Scenes

In this letter, we address the problem of collision avoidance for a swarm of UAVs used for continuous surveillance of an urban environment. Our method, LSwarm, efficiently avoids collisions with static obstacles, dynamic obstacles and other agents in three-dimensional urban environments while considering coverage constraints. LSwarm calculates collision avoiding velocities that maximize the conformity of an agent to an optimal path given by a global coverage strategy and ensure sufficient resolution of the coverage data collected by each agent. Our algorithm is formulated based on optimal reciprocal collision avoidance and is scalable with respect to the size of the swarm. We evaluate the coverage performance of LSwarm in realistic simulations of a swarm of quadrotors in complex urban models. In practice, our approach can compute collision avoiding velocities for a swarm composed of tens to hundreds of agents in a few milliseconds on dense urban scenes consisting of tens of buildings.

Mosquito-inspired distributed swarming and pursuit for cooperative defense against fast intruders

Inspired by the swarming behavior of male mosquitoes that aggregate to attract and subsequently pursue a female mosquito, we study how random swarming motion in autonomous vehicles affects the success of target capture. We consider the scenario in which multiple guardians with limited perceptual range and bounded acceleration are deployed to protect an area from an intruder. The main challenge for the guardian (male mosquito) is to quickly respond to a fast intruder (female) by matching its velocity. We focus on the motion strategy for the guardians before they perceive the intruder, which we call the swarming phase. In the parameter space consisting of the intruder’s speed and guardians’ ability (i.e., maximum acceleration and perceptual range) we identify necessary and sufficient conditions for target capture. We propose a swarming algorithm inspired by the behavior of male mosquitoes to improve the target-capture capability. The theoretical results are illustrated by experiments with an indoor quadrotor swarm.

Motion optimization algorithm designing for swarm quadrotors in application of grasping objects

In this study, the process of designing a motion optimization algorithm for swarm quadrotor robots is presented. Motions equations of swarm are written based on Lagrangian energy equations. A potential function is applied on the equations to optimize the swarm motion. The applied potential function enables each of the swarm members to move toward an independent target coordinate as motion starts and simultaneously connecting with other members. As a result, the necessity of having the members aggregated within an area close to the swarm center is eliminated. This algorithm is supposed to act on swarm of quadrotors; therefore a validated dynamic model of quadrotor and a designed controller are introduced to discuss the possible applications. The designed algorithm is then applied to grasp an object. In order to establish grasping, particle swarm optimization method is used. Finally, the algorithm is simulated in MATLAB for a two-member swarm of quadrotors for grasping the object. Simulation results indicate increased work space for the members along the motion path and reduced mission time.

Agile Coordination and Assistive Collision Avoidance for Quadrotor Swarms Using Virtual Structures

This paper presents a method for controlling a swarm of quadrotors to perform agile interleaved maneuvers while holding a fixed relative formation, or transitioning between different formations. The method prevents collisions within the swarm, as well as between the quadrotors and static obstacles in the environment. The method is built upon the existing notion of a virtual structure, which serves as a framework with which to plan and execute complex interleaved trajectories, and also gives a simple, intuitive interface for a single human operator to control an arbitrarily large aerial swarm in real time. The virtual structure concept is integrated with differential flatness-based feedback control to give an end-to-end integrated swarm teleoperation system. Collision avoidance is achieved by using multiple layered potential fields. Our method is demonstrated in hardware experiments with groups of 3–5 quadrotors teleoperated by a single human operator, and simulations of 200 quadrotors teleoperated by a single human operator.

Trajectory Planning for Quadrotor Swarms

We describe a method for multirobot trajectory planning in known, obstacle-rich environments. We demonstrate our approach on a quadrotor swarm navigating in a warehouse setting. Our method consists of following three stages: 1) roadmap generation that generates sparse roadmaps annotated with possible interrobot collisions; 2) discrete planning that finds valid execution schedules in discrete time and space; 3) continuous refinement that creates smooth trajectories. We account for the downwash effect of quadrotors, allowing safe flight in dense formations. We demonstrate computational efficiency in simulation with up to 200 robots and physical plausibility with an experiment on 32 nano-quadrotors. Our approach can compute safe and smooth trajectories for hundreds of quadrotors in dense environments with obstacles in a few minutes.

Visual Inertial Odometry Swarm: An Autonomous Swarm of Vision-Based Quadrotors

In this letter, we present the system infrastructure for a swarm of quadrotors, which perform all estimation on board using monocular visual inertial odometry. This is a novel system since it does not require an external motion capture system or GPS and is able to execute formation tasks without inter-robot collisions. The swarm can be deployed in nearly any indoor or outdoor scenario and is scalable to higher numbers of robots. We discuss the system architecture, estimation, planning, and control for the multirobot system. The robustness and scalability of the approach is validated in both indoor and outdoor environments with up to 12 quadrotors.

Lattice flocking of multi-quadrotor system: an algorithm based on artificial potential field

This paper investigates the flocking and obstacle avoidance problem of multi-quadrotor system with a lattice-type structure. To tackle nonlinear dynamics and cooperative consensus of a swarm of quadrotors, we proposed two algorithms (one for free flocking and one for constrained flocking) based on artificial potential field with two cooperative strategies using position tracking and attitude tracking, respectively. Firstly, the inner-loop control model of single quadrotor is provided. Then, we build an artificial potential function to act on the coordinate variable. Third, obstacle avoidance term is added to the flocking algorithm proposed previously. Finally, the position and attitude are taken as the coordinate variables, respectively, to both two algorithms, one for optimizing the distance of each quadrotor until the lattice-type structure is formed and keeping continuously and another for optimizing the distance between quadrotor and obstacles to avoid collision. Simulation results demonstrate the validity of both algorithms.

Utilization of the Physicomimetics Framework for Achieving Local, Decentralized, and Emergent Behavior in a Swarm of Quadrotor Unmanned Aerial Vehicles (QUAV)

This paper presents the implementation of the physicomimetics framework in governing the behavior of a swarm of quadrotors. Each quadrotor uses only local information about itself and the neighboring quadrotors to determine its own movement by applying the principles of physicomimetics. Through these localized and relatively simple interactions, the swarm of quadrotors was able to organize itself into various structures and exhibit different swarm behaviors such as aggregation, obstacle avoidance, lattice formation, and dispersion.

Implementation of Swarm Social Foraging Behavior in Unmanned Aerial Vehicle (UAV) Quadrotor Swarm

One of the novel approaches in multiple quadrotor control is swarm robotics. It aims to mimic social behaviors of animals and insects. This paper presents the physical implementation of the swarm social foraging behavior in unmanned aerial vehicle quadrotors. To achieve this, it first explores the basic behavior of aggregation. It is implemented over a quadrotor swarm test-bed that makes use of external motion capture cameras. The completed algorithm makes use of the artificial potential function model combined with the environment resource profile model. Results show successful demonstration of the social foraging algorithm with minimal error in position. Also, the proposed algorithm’s performance presents an increase in aggregation speed and time as the number of swarm member increases.

Smoothed Particle Hydrodynamics Approach to Aggregation of Quadrotor Unmanned Aerial Vehicle Swarm

This paper uses a fluid mechanics approach to perform swarming aggregation on a quadrotor unmanned aerial vehicle (QUAV) swarm platform. This is done by adapting the Smoothed Particle Hydrodynamics (SPH) technique. An algorithm benchmarking is conducted to see how well SPH performs. Simulations of varying set-ups are experimented to compare different algorithms with SPH. The position error of SPH is 30% less than the benchmark algorithm when a target enclosure is introduce. SPH is implemented using Crazyflie quadrotor swarm. The aggregation behavior exhibited successfully in the said platform.

Distributed reactive collision avoidance for a swarm of quadrotors

The large-scale of unmanned aerial vehicle applications has escalated significantly within the last few years, and the current research is slowly hinting at a move from single vehicle applications to multivehicle systems. As the number of agents operating in the same environment grows, conflict detection and resolution becomes one of the most important factors of the autonomous system to ensure the vehicles’ safety throughout the completion of their missions. The work presented in this paper describes the implementation of the novel distributed reactive collision avoidance algorithm proposed in the literature, improved to fit a swarm of quadrotor helicopters. The original method has been extended to function in dense and crowded environments with relevant spatial obstacle constraints and deconfliction manoeuvres for high number of vehicles. Additionally, the collision avoidance is modified to work in conjunction with a dynamic close formation flight scheme. The solution presented to the conflict detection and Resolution problem is reactive and distributed, making it well suited for real-time applications. The final avoidance algorithm is tested on a series of crowded scenarios to test its performances in close quarters.

A Reliable Open-Source System Architecture for the Fast Designing and Prototyping of Autonomous Multi-UAV Systems: Simulation and Experimentation

During the process of design and development of an autonomous Multi-UAV System, two main problems appear. The first one is the difficulty of designing all the modules and behaviors of the aerial multi-robot system. The second one is the difficulty of having an autonomous prototype of the system for the developers that allows to test the performance of each module even in an early stage of the project. These two problems motivate this paper. A multipurpose system architecture for autonomous multi-UAV platforms is presented. This versatile system architecture can be used by the system designers as a template when developing their own systems. The proposed system architecture is general enough to be used in a wide range of applications, as demonstrated in the paper. This system architecture aims to be a reference for all designers. Additionally, to allow for the fast prototyping of autonomous multi-aerial systems, an Open Source framework based on the previously defined system architecture is introduced. It allows developers to have a flight proven multi-aerial system ready to use, so that they can test their algorithms even in an early stage of the project. The implementation of this framework, introduced in the paper with the name of CVG Quadrotor Swarm, which has also the advantages of being modular and compatible with different aerial platforms, can be found at https://github.com/Vision4UAV/cvg_quadrotor_swarm with a consistent catalog of available modules. The good performance of this framework is demonstrated in the paper by choosing a basic instance of it and carrying out simulation and experimental tests whose results are summarized and discussed in this paper.