mjhaskell / roscopter

*Under Development* - A fully-featured multirotor autopilot for ROS, based on Dr. Beard's Quadrotor Notes

ROScopter

This repository contains the ROS packages associated with flying a multirotor on the ROSflight autopilot stack. The roscopter ROS package contains a high-level multirotor autopilot while the roscopter_sim package is used for simulation.

The roscopter autopilot has been built with reference to Dr. Randy Beard's Quadrotor Dynamics and Control notes, so as to allow anyone to easily understand, modify and use the code. The framework developed in the aforementioned reference closely resembles the fixed-wing framework developed in Small Unmanned Aircraft by Beard and McLain. This framework is inherently modular and extensively documented so as to aid the user in understanding and extending for personal use.

Running Simulation

Simplified Dynamics Model

Clone this repo and the ROSflight_plugins repo into catkin_ws/src, you can run the Gazebo simulator with

$ catkin_make
$ source devel/setup.bash
$ roslaunch roscopter_sim multirotor.launch

This simulation does not simulate the firmware and does not require a joystick. It has the same API as the SIL stack, and can be useful for developing high-level algorithms, however it makes some simplifying assumptions in the dynamics which reduces the fidelity of the simulation.

Software In The Loop

To run the rosflight SIL simulation information with the full roscopter stack, clone the ROSflight repo, connect a joystick or transmitter and run

$ catkin_make
$ source devel/setup.bash
$ roslaunch roscopter copter_sil..launch

Core Autopilot Components

EKF

An Extended Kalman Filter (EKF) on manifold is used for state estimation using ROSflight attitude estimates as measurements. See Derivation of the Relative Multiplicative Kalman Filter by David Wheeler and Daniel Koch for more information.

The EKF supports measurements from a position and attitude estimator (motion capture system), altitude sensor (sonar or barometer) and IMU. GPS support is currently being developed. Measurement models can be added quite easily to the filter, please create an Issue if you have a measurement model you want supported and we can walk you through the steps on how to do it.

Controller

A PID controller is used.

Waypoint Manager

Velocity manager and commander.