Data-Driven System Identification of Quadrotors Subject to Motor Delays

Jonas Eschmann,Dario Albani,Giuseppe Loianno

2024-09-24

Abstract:Recently non-linear control methods like Model Predictive Control (MPC) and Reinforcement Learning (RL) have attracted increased interest in the quadrotor control community. In contrast to classic control methods like cascaded PID controllers, MPC and RL heavily rely on an accurate model of the system dynamics. The process of quadrotor system identification is notoriously tedious and is often pursued with additional equipment like a thrust stand. Furthermore, low-level details like motor delays which are crucial for accurate end-to-end control are often neglected. In this work, we introduce a data-driven method to identify a quadrotor's inertia parameters, thrust curves, torque coefficients, and first-order motor delay purely based on proprioceptive data. The estimation of the motor delay is particularly challenging as usually, the RPMs can not be measured. We derive a Maximum A Posteriori (MAP)-based method to estimate the latent time constant. Our approach only requires about a minute of flying data that can be collected without any additional equipment and usually consists of three simple maneuvers. Experimental results demonstrate the ability of our method to accurately recover the parameters of multiple quadrotors. It also facilitates the deployment of RL-based, end-to-end quadrotor control of a large quadrotor under harsh, outdoor conditions.

Robotics,Systems and Control

What problem does this paper attempt to address?

The problems that this paper attempts to solve are the challenges in the system identification process of quadrotor drones, especially for the application of nonlinear control methods such as model predictive control (MPC) and reinforcement learning (RL). Specifically, the paper focuses on the following aspects:

1. **Requirement for accurate dynamic models**: Unlike traditional cascaded PID controllers, modern control methods such as MPC and RL are highly dependent on accurate dynamic models of the system. However, the current quadrotor system identification process is complex and time - consuming, and usually requires additional equipment (such as a thrust stand) to obtain accurate parameters.

2. **Neglect of low - level details**: Existing system identification methods often overlook low - level details that are crucial for end - to - end control, such as motor delays. These details are very important for achieving high - precision control but are usually difficult to measure.

3. **Data - driven method**: The paper proposes a data - driven method, aiming to identify the inertial parameters, thrust curves, torque coefficients, and first - order motor delay time constants of the quadrotor using only proprioceptive data (such as accelerometers, gyroscopes, and motor commands) and simple measurements (such as mass, motor position).

4. **Simplifying the system identification process**: The goal of the paper is to provide a simple and robust system identification method, making the burden of this process lower than the tuning burden of traditional PID controllers, thereby promoting the wide application of nonlinear control methods on multiple platforms.

By solving these problems, the paper hopes to accelerate the research on quadrotor drones and promote the application of more advanced control methods in industry.