DESIGN & OPTIMIZATION OF PROGRAMMABLE MOTION PROFILING & CONTROL WITH AUTONOMOUS UAV APPLICATIONS

Yu, Wenjie

Permalink

https://hdl.handle.net/2142/124902 Copy

Description

Title

DESIGN & OPTIMIZATION OF PROGRAMMABLE MOTION PROFILING & CONTROL WITH AUTONOMOUS UAV APPLICATIONS

Author(s)

Yu, Wenjie

Issue Date

2021-05-01

Keyword(s)

S-curve, Higher Order Motion Profiling, Jerk, Motion Control, UAV, Quadrotor, Pixhawk, Ardupilot, MavLINK, DroneKit

Abstract

Many of the motion controllers in the industry use non-programmable motion profiles, which are often limited to modeling a trapezoidal profile with a linear acceleration/deceleration phase, or an s-curve profile of jerk, the derivative of acceleration. Higher order derivatives of velocity are rarely explored to achieve a customized smoothness of the motion profile, which is desired in many real-world applications and robotics systems. The lack of programmability and open-source platforms makes the price of customizing control expensive and the design process redundant for different applications. This thesis provides a low-cost, open-source, and programmable solution that generates motion profiles for a wide range of customized precision levels. We first summarize the common assumptions and generic constraints required to create a smooth motion profile. We review existing approaches of velocity profiling to show the challenges in providing programmability, then offer two implementations of the feedforward motion profiling system as our solution. Performance evaluations of the two implementations are done both mathematically and empirically. Then, this thesis offers an overview on how our motion profile methodology can be integrated into a closed-loop control system of any kind with existing open-source PID feedback for some specific applications. We will use a 1D translation rail to illustrate the effectiveness of higher order motion profiling and the quadrotor demo for autonomous flight to show the compatibility and programmability of the algorithm in large complex systems.

Type of Resource

text

Language

eng

Owning Collections