To isolate the algorithm performance in AI Challenge, the competition committee restricted the robot models since ICRA2019 (for more details, please click). Now in Team Ausdroid, two sets of standard robotic platforms are available in Doug Mcdonell Building. Both robots were modified to carry the camera, Lidar and computing device (J120 or Manifold 2). Except regular maintenance, not much fancy work needs to be developed in structure. But there are some important development tasks in embedded programming. Some potential projects in Hardware group beside routine maintenance are listed below. (NOTE: Considering the current COVID-19 situation, the robotic devices are no longer accessible until further notice. The available projects at this stage are marked with *.)


In past two years, three different types of mini computing devices which share the same core, TX2, were installed into the robots. In this year, with the generous support from CIS, computing devices for both robots were upgraded to NVIDIA Jetson AGX Xavier which is stronger in computing but bigger in dimension. This task requires

  1. Re-organize the internal space of the robot, including re-wiring.
  2. Design the mounting part for Xavier and install it inside the robot firmly.*
  3. Install other stuff inside the robot (if needed).
  4. Design a set of cooling system to guarantee the performance of the computing device.

Camera and Lidar Installation

As introduced above, the camera and Lidar are well installed on the robot. However there are still some minor points can be imporved. For example,

  1. Convenience of disassembly and assembly.
  2. Convenience of adjusting camera configuration (aperture and focal length).

Embedded Programming

The hadrware controllers embedded in RoboMaster AI robot are two pieces of RoboMaster Development Board Type A (STM32). The source code for embedded system is open source on GitHub. Recently, patrol function (gimbal rotates slowly to search enemy while the chassis works normally) has been added and can be called by switching left toggle on the remote controller. The further tasks are

  1. Develop the dodge function (chassis rotates quickly to avoid attacking while gimbal keeps stable to attack enemy).*
  2. Develop communication protocols for both functions to enable automatic control.*

If interested, what's next?

The following documents might be useful for you to explore more about the robotic platform. For more information, please contact Murphy (y.xu100@student.unimelb.edu.au).