Design, development and implementation of swarm behaviors for quadrotor unmanned aerial vehicle (QUAV) swarm

• Main Author: Bandala, Argel A.
• Format: text
• Language: English
• Published: Animo Repository 2015
• Online Access: https://animorepository.dlsu.edu.ph/etd_doctoral/429
• Institution: De La Salle University

Different variants of robots are made known to humans. Mobile robots or mobots were discovered a few years back and became an interest for technologists and engineers. These robots introduced adaptability and versatility because of its fundamental characteristic of being mobile. Amongst the type of mobile robots, aerial vehicles are the least explored. Ironically, aerial robots offer more versatility and mobility as compared to other types of mobile robots, such as land and water. As compared to its counterparts, aerial vehicles offer a wide range of applications for its characteristic of larger work envelope. Furthermore, the abilities and capabilities of the robots are reliant to the behaviours and control algorithms embedded to it. In recent years, swarm intelligence gained popularity because of its inherent characteristic of being scalable and independent. However, there are very few instances of using these algorithm in mobile robots. The exploration of distributed robotics in research grew because of the numerous applications that can be unlocked upon its perfection. This study is initiated to develop and apply swarming behaviors derived from natural inspirations such as social animals and insects in a swarm of quadrotor unmanned aerial vehicles. The swarming behaviors are expected to manifest in physical behaviors and movements of the robots. Certain tasks will be the platform for testing of the applicability and success of the developed behavior algorithms. Three swarm behaviors were explored in this study, namely, aggregation, foraging and formation. The researcher explored two methods to realize the said behaviors. The first method is the derived swarm behaviors from two dimensional swarm methods. The second method is by approximating swarm members as fluid particles and using smoothed particle hydrodynamics method. The testing platform is consists of nano quadrotors which will serve as the boids. These quadrotors are equipped with orientation sensors such as accelerometer and gyroscope. The quadrotors are also equipped with pressure sensor which generates the altitude reading of the robot. Two microcontrollers are utilized by the robot to execute and facility wireless communication, sensor fusion and motor control. Another part of the testing platform is consists of external observer which is realized by using a motion capture system. The motion capture system is composed of cameras which detects infrared light. The robots are equipped with reflectors of infrared light which is then detected by the motion capture system. The aggregation behavior in SPH using plane container yielded 84.88%. On the other hand using a spherical container, the accuracy is around 95.23%. Lastly using a toroidal container, the accuracy produced is 92.44%. Increasing the number of swarm elements or particles decreases the accuracy or aggregation of the elements in the aggregation point. Since every element is constrained by the container, convergence at the gravity point is not possible. Furthermore, increasing the number of particles will generate higher sum of particle distance from the point of aggregation. The reason for this is that a particle occupies space in the container and would likely add up to the distance from the gravity source of other particles.