Things to do during summer school

Overall task:

The teams will go through a search and rescue operation where they will use a team of vehicles consisting of a ground vehicle (Turtlebot) and UAVs (ARDrone) to find objects in a remote area that cannot be accessed by you, only the vehicles. This is a human-in-the-loop setup where a human can perform tasks that you are not able to perform autonomously.

The area will contain a set of different types of objects like QR Codes, Boxes, shapes (square, circle etc). Some of these will be easy to spot, some can only be spotted from above and some only from the ground.

The vehicles will operate in one area and the human operators will be in another area. The only way to communicate between the humans and the vehicles is via a network connection. In principle the distance between humans and vehicles could be kilometers, but in the summer school setup it will be tens of meters but without line of sight.

Your team will scored based on the two criteria; i) the autonomy of the task and ii) the objects spotted and your ability to localize them.

Tasks – Day by Day:

  1. Getting started
    Get used to Turtlebot and ARDrone and Teleoperate both platforms. This will be the baseline for your system.
    In the links bellow, you can find tutorials for the first task. The access to the tutorials is in suggest mode, please    post all your suggestions to improve them in the form of comments, do not edit them. We will fix them asap.
    If you have any queries, please contact us on slack 16_summerschool channel or in a direct message to    Pradeep Korivi or Vladimir Li.
    Turtlebot tutorial here!
    Ardrone turtorial here!
  2. Navigation:
    To getting started you should download packages from git following this tutorial.

    1. Make turtlebot move autonomously using out of the box gmapping and exploration algorithms. Tutorial!
    2. Control the drone. Tutorial!
  3. Perception
    1. Object detection and recognition in 2D and OpenCV.
    2. Object detection and recognition in 3D and PCL.
  4. System integration and develop strategies for competition and finalize.
  5. Competition Day: Final Adjustments and competition.

 

Additional Information:

Connection to core topics in WASP

  • Control and decision making:
    • Your need to control the platforms. Thsi will be relatively simple for the ground vehicle here and you will likely be able to reach quite high levels of autonomy in this aspect. For the drone things will be quite a lot more complicated and we expect that you will have to use more human control.
    • Decision making will also be very important. In your baseline system all of the decision making will be done by humans but we hope that some of you will be able to achieve some level of autonomy here when it comes to for example deciding on how to explore the environment.
  • Sensing and perception:
    • Computer vision for objects detection, identification and position will be central to solving the task with high level of autonomy.
    • Localization and mapping for the ground vehicle will allow for higher levels of autonomy.
  • Learning and knowledge:
    • You will definitely identify the need for learning object models, performing repeated tasks, etc but we expect that it will be out of scope for the summer school to actually use learning.
    • Most of the knowledge will be handled by the humans in this tasks but you will clearly see the need to encode and exploit knowledge to solve tasks more efficiently and to store information/experience that you have gained when performing tasks.
  • Interaction and collaboration:
    • This will be the essence of the task. Interaction and collaboration between platforms and between platforms and humans. Visualization will also be there although in very primitive forms but it will probably be quite clear that good ways to visualize information will be very helpful.
  • System and Software:
    • It will be very clear to everyone that a good understanding of systems and software will be important. During the summer school the task will have a robotics flavour but autonomy in most forms has to be realised with software.