Home

Welcome to the ethz-asl/mav_dji_ros_interface Wiki!

This repository contains a modified version of DJI Onboard ROS SDK (3.2) designed to give greater control over the DJI development platforms. We hope our framework helps you and please remember to cite us in your work!

Brief description

This repository presents a complete visual-inertial (VI-) odometry-aided MAV platform that makes use of off-the-shelf elements such as the device itself, a computer, and VI-sensor. In addition, complete documentation is provided which records every step taken to build the MAV. The fundamental idea underlying this project is to contribute to the robotics community by releasing the knowledge and skills that we have learned while running this project. Therefore, this project is completely open-source; including software packages, 3D models, and all parts.

Using our tools, researchers can perform their own system identification without a motion capture system. This is important to achieve decent control performance in case some property of the system has changed (e.g., the center of mass or the moment of inertia). An example dataset and code are provided as guidance on how to do this experiment. The identified dynamics model is utilized within a nonlinear MPC framework and pose-tracking errors prove that the estimated models are accurate.

Moreover, this package provides a camera and multi-IMU calibration (intrinsic and extrinsic) example using a commercial VI-sensor and MAV. Visual-odometry estimates making use of the calibration validate that our calibration is precise and correct.

We wish to hear about your feedback in order to improve this project and fix some possible bugs. Please let us know about your developments! - from the Flourish and MAV team at ETH ASL.

Use the sidebar on the right to navigate through the Wiki.

Publications

If our work helps your works in an academic/research context, please cite the following publication(s):

I. Sa, M. Kamel, R. Khanna, M. Popovic, J. Nieto and R. Siegwart, "Dynamic System Identification, and Control for a cost-effective open-source VTOL MAV", 2017, Field of Service Robotics. (PDF)

@ARTICLE{2017M100Ctrl,
          author = {{Sa}, I. and {Kamel}, M. and {Khanna}, R. and {Popovic}, M. and {Nieto}, J. {Siegwart}, R.},
          title = "{Dynamic System Identification, and Control for a cost effective open-source VTOL MAV}",
          archivePrefix = "arXiv",
          eprint = {1701.08623},
          primaryClass = "cs.RO",
          keywords = {Computer Science - Robotics},
          year = 2017,
          month = Jan
}

I. Sa, M. Kamel, M. Burri, M. Bloesch, R. Khanna, M. Popovic, J. Nieto, and R. Siegwart, "Build your own visual-inertial odometry aided cost-effective and open-source autonomous drone", 2017, IEEE Robotics & Automation Magazine. (PDF)

@article{sa2017build,
  title={Build your own visual-inertial odometry aided cost-effective and open-source autonomous drone},
  author={Sa, Inkyu and Kamel, Mina and Burri, Michael and Bloesch, Michael and Khanna, Raghav and Popovic, Marija and Nieto, Juan and Siegwart, Roland},
  journal={arXiv preprint arXiv:1708.06652},
  year={2017}
}