Claim Missing Document
Check
Articles

Found 2 Documents
Search

Swarm control of an unmanned quadrotor model with LQR weighting matrix optimization using genetic algorithm Joelianto, Endra; Christian, Daniel; Samsi, Agus
Journal of Mechatronics, Electrical Power and Vehicular Technology Vol 11, No 1 (2020)
Publisher : Research Centre for Electrical Power and Mechatronics, Indonesian Istitutes of Sciences

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.14203/j.mev.2020.v11.1-10

Abstract

Unmanned aerial vehicle (UAV) quadrotors have developed rapidly and continue to advance together with the development of new supporting technologies. However, the use of one quadrotor has many obstacles and compromises the ability of a UAV to complete complex missions that require the cooperation of more than one quadrotor. In nature, one interesting phenomenon is the behaviour of several organisms to always move in flocks (swarm), which allows them to find food more quickly and sustain life compared with when they move independently. In this paper, the swarm behaviour is applied to drive a system consisting of six UAV quadrotors as agents for flocking while tracking a swarm trajectory. The swarm control system is expected to minimize the objective function of the energy used and tracking errors. The considered swarm control system consists of two levels. The first higher level is a proportional – derivative type controller that produces the swarm trajectory to be followed by UAV quadrotor agents in swarming. In the second lower level, a linear quadratic regulator (LQR) is used by each UAV quadrotor agent to follow a tracking path well with the minimal objective function. A genetic algorithm is applied to find the optimal LQR weighting matrices as it is able to solve complex optimization problems. Simulation results indicate that the quadrotors' tracking performance improved by 36.00 %, whereas their swarming performance improved by 17.17 %.
Sistem Kontrol Swarm untuk Flocking Wahana NR-Awak Quadrotor dengan Optimasi Algoritma Genetik Joelianto, Endra; Rajasa, Winarendra Satya; Samsi, Agus
Jurnal Teknologi Informasi dan Ilmu Komputer Vol 8 No 6: Desember 2021
Publisher : Fakultas Ilmu Komputer, Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.25126/jtiik.2021863467

Abstract

Quadrotor merupakan wahana udara nir-awak jenis lepas landas atau pendaratan vertikal berbentuk silang dan memiliki sebuah rotor pada setiap ujung lengannya dengan kemampuan manuver yang tinggi. Swarm quadrotor yang terdiri dari sekumpulan quadrotor akan menjadi suatu swarm yang baik, sesuai dengan kriteria swarm oleh Reynold yaitu dapat menghindari tumbukan, menyamakan kecepatan, dan pemusatan swarm. Pengontrolan swarm quadrotor memiliki tingkat kerumitan yang tinggi karena melibatkan banyak agen. Riset pengembangan swarm quadrotor masih belum banyak dilakukan dan masih membuka peluang untuk meneliti dengan metoda lain yang lebih baik dalam menghasilkan swarm. Makalah ini mengusulkan pengontrolan swarm quadrotor yang terdiri dari dua tingkat lup kontrol. Lup pertama adalah pengontrol sistem model swarm untuk membangkitkan lintasan swarm dan lup kedua merupakan pengontrol pada quadrotor untuk melakukan penjejakan lintasan swarm. Pengontrol pertama menggunakan pengontrol proporsional derivatif (PD), sedangkan pengontrol kedua menggunakan regulator linier kuadratik (RLK). Pengontrol yang dirancang memiliki parameter yang banyak, sehingga pemilihan parameter yang optimal sangat sulit. Pencarian parameter optimal pada pengontrol model swarm quadrotor membutuhkan teknik optimasi seperti algoritma genetik (AG) untuk mengarahkan pencarian menuju solusi yang menghasilkan kinerja terbaik. Pada makalah ini, penalaan dengan optimasi AG hanya dilakukan pada pengontrol PD untuk menghasilkan lintasan swarm terbaik, sedangkan matrik bobot RLK dilakukan secara uji coba. Hasil simulasi swarm pada model quadrotor menunjukkan parameter , . , dan  yang diperoleh menggunakan AG menghasilkan pergerakan swarm yang baik dengan kesalahan RMS pelacakan 0,0094 m terhadap fungsi obyektif. Sedangkan ketika parameter ,  dan  dicari menggunakan AG, tidak berpengaruh banyak dalam memperbaiki hasil simulasi swarm quadrotor. AbstractThe quadrotor is a type of take-off or vertical landing unmanned aerial vehicles with a cross shape and has one rotor at each end of its arm with high maneuverability. A quadrotor swarm consisting of a group of quadrotors leads to a good swarm, according to Reynold's swarm criteria, which accomplishes collision avoidance, velocity matching, and flock centering. Quadrotor swarm control has a high level of complexity because it involves many agents. Research on the development of quadrotor swarm has received insignificant attention and it still opens opportunities to research other methods that are better at producing swarm. The paper proposes the control of a quadrotor swarm consisted of two levels of control loops. The first loop controls the swarm model system to generate the swarm trajectory and the second loop is the controller on the quadrotor to track the swarm path. The first controller uses a proportional derivative controller (PD), while the second controller uses the linear quadratic regulator (LQR). The controller that is designed has many parameters, so the optimal parameter selection is very difficult. The search for optimal parameters in the swarm model controller requires optimization techniques such as the genetic algorithm (GA) to direct the search for solutions that produce the best performance. In this paper, tuning with the optimization of GA is only done for the PD controller in order to produce the best swarm trajectory, while the weight matrices of the LQR are done on a trial error basis. Swarm simulation results of a quadrotor model system show the parameters , . , and  obtained using GA produce a good swarm movement with RMS error 0.0094 m of the objective function. Whereas when parameters ,  and  are searched using GA, it does not have much effect in improving the quadrotor swarm simulation results.