Improved Distributed Formation Control and Trajectory Tracking of Multi Quadrotor in Leader-Follower Formation Billie Pratama 1 , Abdul Muis 2 , Aries Subiantoro 3 Electrical Engineering Department Universitas Indonesia, Indonesia 1 billie.pratama@ui.ac.id, 2 amuis@ieee.org, 3 aries.subiantoro@ui.ac.id Abstract—This paper focuses on two quadrotors with leader- follower formation to conduct a formation tracking. This paper proposes simplified control for this formation tracking. Each quadrotor has two controllers; position and orientation controller. There are only four control parameters that need to be tuned properly. Most of the formation controls are based on the position reference. In general, the follower uses the leader position as its position reference. In this paper, a distributed formation control based on velocity reference is developed. The state of each quadrotor is distributed and the follower uses the velocity command of leader to obtain the velocity signal. The algorithms are validated through simulations and real-time implementations. The results show that the controllers realize the formation tracking. Keywords—Multi quadrotor, ROS, formation control, PI controller, trajectory tracking I. INTRODUCTION Nowadays, quadrotor has remarkable potentials in various fields, like transportations, commercials, and militaries. Moreover, a quadrotor with the addition of camera can be utilized more optimally, such as for regional monitoring[1], assessment of agricultural crops [2], and aerial photography [3]. Many complicated tasks cannot be handled by single quadrotor. A multi quadrotor system is required to enhance the efficiency, extend the view, and improve the effectiveness. Each quadrotor in the system must achieve good coordination for completing a mission. The coordination of multi quadrotor system is commonly inspired by biological formation behaviours of natural world. Recently, multi quadrotor system has attracted research interest and applied in several areas. In [4], the control of multiple quadrotor was designed for persistent surveillance. By using several quadrotors, the monitoring view can be expanded and increase the effectiveness. Moreover, multi quadrotor system is lately observed for transport things, search and rescue. In general, the control schema of the formation has three types: centralized, decentralized, and distributed control strategies [5]. Centralized control [6] is a simple and easy implementing where there is one central system that collects the data of other subsystems and makes decisions for them. Nevertheless, centralized control has bad robustness as it is susceptible to the fault of central unit. To deal with the robustness problem, a decentralized control [7] was proposed by Koksal et.al. In which, each quadrotor has independent controllers. Unfortunately, a predefined path to all quadrotors should be provided. Also there is no distributed information among them. To overcome this problem, this paper only provides predefined path only to leader and then distributes its states to followers. The follower is controlled to follow leader state. Here, the control problem is simplified and can be scaled to more followers easily. Even though the multi quadrotor system has so many advantages, there are still a lot of considerations and problems to deal with. Each quadrotor in the system must be controlled properly, either as a single quadrotor or as a team. There are already some studies related to the control system of multi quadrotor. In [8], Hao Liu et al concerned with the formation control problem of a multi quadrotor system to achieve aggressive trajectory tracking with prescribed formation patterns. The controller is designed by Linear Quadratic Regulator (LQR) based approach. D. A. Mercado et al [9] present quadrotors flight formation control using a leader- follower approach. A sliding mode controller is used on the outer loop and provides the desired attitude for the inner loop, which is controlled by a PD controller. Most of the control system have many control parameters to be properly tuned. In fact, the control performances depend on the value of these parameters. In this paper, the formation tracking control is improved by simplifying the control system. In which, only four control parameters are tuned; three belong to leader and one belong to follower. In addition, a formation control based on velocity reference is developed in this paper. The follower uses the velocity reference of the leader in tracking the trajectory. As a result, the leader and the follower accelerate at the same time and the position errors between them can be minimized significantly. The remaining portion of this paper is organized as follows: Section II explains the dynamic system of quadrotor, section III explains the path planning and control system for