Home

Welcome to the TrajBridge wiki. Here we describe the entire setup of getting a multirotor to fly autonomously using the pairing of two popular computation platforms in research:

1. A PX4 flight controller for low-level control
2. An Ubuntu companion computer (x86 or ARM)

Of specific interest to us, and the focus of this repository, is the amalgamation of several existing research tools:

1. A hardware robust multirotor frame with good payload capacity and flight time
2. A motion capture 'ground truth' reference
3. A simple low fidelity, physics only simulation
4. An accessible high fidelity simulation that is tightly linked to the actual flight setup
5. Access to low level control inputs for agile flight
6. Easy implementation of learning-based policies with minimal dependency on the rest of the infrastructure
7. Easy testing in hardware.

Many of these things already exist independently and are mature ongoing developments, but putting them all together has been... quite the experience. So really, this repository is an attempt to very clearly explain how to put them all together.

The wiki is structured as follows:
Setup Overview: Description of the hardware and software we use and how they are connected to each other.
Simulation: Setting up a software-in-the-loop simulation. All you need is a laptop!
Drone Hardware: Assembling a good research drone. This is a deep dive on strictly the hardware.
Flight Controller: Configuring the PX4 to link it to a high level controller and motion capture.
Companion Computer: The high level controller where we will be implementing research stuff.
External Pose Estimate: Here we implement a ground truth reference that uses motion capture.
MSL Drone Tracker: Updated list of current MSL drones.\

Given that many of the tools that we use are products of ongoing development, we felt that it was prudent to fork and 'freeze' many of the repositories that we use. Some of these are internal to StanfordMSL. Contact one of the repository contributors if you want access.