Welcome to ROSflight¶
ROSflight is an autopilot system designed from the ground up with researchers in mind, and for easy integration with ROS. Its purpose is to accomplish two main tasks:
- Stream high-rate sensor data from the embedded flight controller to ROS
- Provide a simple API for sending control setpoints to the embedded flight controller
A ROSflight setup consists of two main components:
- The embedded flight controller: Typically an OpenPilot Revo or one of its variants, running the ROSflight firmware
- The companion computer: Any computer that runs ROS, is located on the vehicle, and has a physical serial connection to the embedded flight controller
There are a lot of excellent autopilots out there with a lot of great firmware options. Why did we feel like the world needed ROSflight? Because in our experience none of the other available options satisfied our research needs. Specifically, we needed an autopilot that could stream sensor data at high rates, easily accept control setpoints from a companion computer, and accomplish all of this with a lean, easy-to-understand code base.
The other options that we tried were limited in bandwidth for streaming sensor data, and the APIs for sending control setpoints were confusing and difficult to implement. Perhaps most importantly, the code was sometimes so complex (feature-rich, but complicated) that it was difficult to figure out what the autopilot was actually doing. In talking to other researchers and industry members, we found that many people shared similar frustrations. So we decided to create and share the autopilot we wanted, hoping it will be useful to other people as well.
Perhaps more important than what we are trying to accomplish with ROSflight is what we are not trying to accomplish. This is not intended to be, out-of-the-box, a fully-featured autopilot for high-level tasks such as autonomous GPS waypoint following. There are many good autopilots out there that already do this. Instead, ROSflight is intended to provide the minimal functionality required to keep a multirotor or fixed-wing vehicle in the air, and to serve as a building block for writing new code to perform these higher-level tasks. (The ROSplane and ROScopter projects are excellent examples of what can be accomplished by building on top of the ROSflight architecture.)
Therefore, one of the primary objectives of the ROSflight autopilot is to avoid feature creep and remain lean. We hope that others will extend our code and build on top of it, and would love to hear about your successes. But for the most part, we will not be incorporating these new features back into the main project. Instead, we hope that ROSflight will remain a lean, core code base that will continue to serve as a launch pad for exciting new projects and applications.