A Minimalistic 3D Self-Organized UAV Flocking Approach for Desert Exploration

Journal of Intelligent & Robotic Systems (2024) 110:75 https://doi.org/10.1007/s10846-024-02108-0 REGULAR PAPER A Minimalistic 3D Self-Organized UAV Flocking Approach for Desert Exploration Thulio Amorim1 · Tiago Nascimento1,2 · Akash Chaudhary2 · Eliseo Ferrante3 · Martin Saska2 Received: 25 August 2022 / Accepted: 30 April 2024 © The Author(s) 2024 Abstract In this work, we propose a minimalistic swarm flocking approach for multirotor unmanned aerial vehicles (UAVs). Our approach allows the swarm to achieve cohesively and aligned flocking (collective motion), in a random direction, without externally provided directional information exchange (alignment control). The method relies on minimalistic sensory requirements as it uses only the relative range and bearing of swarm agents in local proximity obtained through onboard sensors on the UAV. Thus, our method is able to stabilize and control the flock of a general shape above a steep terrain without any explicit communication between swarm members. To implement proximal control in a three-dimensional manner, the Lennard-Jones potential function is used to maintain cohesiveness and avoid collisions between robots. The performance of the proposed approach was tested in real-world conditions by experiments with a team of nine UAVs. Experiments also present the usage of our approach on UAVs that are independent of external positioning systems such as the Global Navigation Satellite System (GNSS). Relying only on a relative visual localization through the ultraviolet direction and ranging (UVDAR) system, previously proposed by our group, the experiments verify that our system can be applied in GNSS-denied environments. The degree achieved of alignment and cohesiveness was evaluated using the metrics of order and steady-state value. Keywords Unmanned aerial vehicles · Flocking · Swarm robotics · Self-organization 1 Introduction Flocking is a coordinated motion of a large group of individuals moving together towards the same target direction (see Fig. 1). This collective behavior can be observed in different species in nature and emulating it in artificial systems has been an active research topic inspired by the fact of how easily simple individuals with limited resources can achieve complex organizations. In robotics, different methods have been proposed to accomplish the flocking behavior in multirobot systems. Those works can be distinguished based on B Tiago Nascimento ; http://mrs.felk.cvut.cz/ Thulio Amorim 1 Department of Computer Systems, Universidade Federal da Paraíba, João Pessoa, Brazil 2 Cybernetics Department, Czech Technical University in Prague, Prague, Czech Republic 3 Technology Innovation Institute, Abu Dhabi, United Arab Emirates particular properties, such as the specification of the robot used as an individual and the environment in which the system can be applied. For example, numerous methods have already been proposed for ground robots acting in a plain environment with external global localization [9]. To accomplish the flocking behavior in the three dimensional (3D) space using flying robots, designing methods with the same desirable attributes, such as scalability, robustness, and flexibility, is still an open problem. Add to that also the need to cope with the limitations on state estimation, such as no use of global information that would restrict the operational space of designed swarming systems. In most state-of-the-art solutions, robot flocking is performed by combining two main control techniques: proximal control and alignment control [4]. The proximal control uses the range and bearing of the neighboring agents to maintain the distance needed to achieve a safe flocking behavior. There is no standard sensor available to obtain the range and bearing of the neighbors of an individual, and most of the swarming systems rely on wireless swarming of states provided by external localization systems. This assumption is not realistic for real-world applications and for large swarms due to 0123456789().: V,-vol 123 75 Page 2 of 14 Journal of Intelligent & Robotic Systems (2024) 110:75 Finally, we can list our contributions as: Fig. 1 Swarm of unmanned aerial vehicles in the desert using our robotic platform the known problems with scalability. Ideally, the individuals should rely on their sensory capabilities to estimate the relative position of their neighbors in local proximity, as it can be observed in nature. In addition, the alignment control uses the orientation of the neighbors to move in a common direction. The usage of an alignment control helps to achieve the flocking more rapidly, but it requires even more elaborate sensing mechanisms to estimate the relative orientation. Different from formation control where more elaborate rules are designed to achieve a complex behavior [19], here a complex behavior is the result of local interactions between individuals through basic rules. According to [4], there is still no formal or precise way to design a collective behavior. Flocking is usually based on virtual physics-based design. This approach draws inspiration from physics where each robot is considered a virtual particle that exerts virtual forces on the other robots within the group. In our work, we apply the method published in [11] for usage in swarms. Thus, we focus on a minimalistic distributed approach for controlling a group of uncrewed aerial vehicles (UAVs) a continuation of work in [13]. Thus, we performed several simulations with ten UAVs and several experimental runs with nine UAVs. The UAVs have embedded control systems in which each UAV is capable of flight control and self-localization. The proposed approach presents a flocking control function that uses only the proximal term (P) in order to converge and move the UAVs into a unified random direction. In this scenario, we can prove that a single function can maintain cohesiveness and hold the flocking orientation unified while moving the flock of UAVs. As a real-world application, we chose to experiment with a group of f450sized UAVs in a desert environment that constantly inserts disturbances into the formation through the difference in terrain level along the flight field. 123 1. a 3D cohesive flocking method based only on proximal control that relies only on relative neighbor position measurement, suitable to environments with remote sensing or radio communication constraints, and irregular terrains; 2. the proposed method has been also integrated with our double-layer control architecture in a manner that imitates non-holonomic behavior, which is needed for proximal control-based flocking, in real-world conditions and where each agent in the flocking uses an onboard localization method; 3. to the best of our knowledge, this is the first time a nine UAV flocking experiment in a real-world environment is performed, in which each UAV has its own onboard control and localization pipeline without an (...truncated)