Initial research platform. Pixhawk and Raspberry Pi.
Initial research platform. Pixhawk and Raspberry Pi.

Research A major part of my Masters by Research at the Bristol Robotics Laboratory is the research project itself.  I am developing unmanned aerial vehicles (UAVs) with the following capabilities:

  1. Able to fly autonomously in a confined space and to map that space.
  2. Able to join with others in a swarm to map the space about them.
  3. Able to identify and locate others in the swarm by:
    1. recognising broadcast ID signals;
    2. using other means to recognise and identify other vehicles (e.g. image recognition).
  4. The objective is that these should eventually be fixed wing, rather than multi-rotor.

This work fits with my interests in the use of UAVs for search and rescue, both on a small scale (think searching unsafe buildings for victims of an earthquake) or on a larger scale, outside, for coordinating search and rescue attempts using swarms. Novel Challenges There is a rapidly growing body of commercial and academic research in this area.  In particular, this project has the following specific challenges:

  1. Not using GPS for navigation
  2. The recognition of other vehicles in a swarm without the use of broadcast signals
  3. ‘Collegiate’, swarm mapping
  4. Flying fixed wing in a confined space

Each of these challenges is significant in itself, and so a step-wise approach is being undertaken, recognising that some elements may simply be beyond the scope of the Masters or even current technologies at this time. Project Approach

  1. Build research platform (test-bed UAV)
  2. Replicate single UAV mapping of environment (others have already done this)
  3. Develop mapping using multiple UAVs (with broadcast IDs/locations)
  4. Develop ability for UAVs to ID and locate each other independently
  5. Miniaturise to fixed wing platform.

This begins to break down the elements into more achievable stages, but of course these may need to be adjusted as the project progresses. Research Platform Flight Control Software With previous experience of hobby UAVs and flight control software, it seems appropriate to build on one of several open-source platforms already available for flight control.  The two major systems under consideration are:

  1. ArduPilot/Copter (
  2. Pixhawk (

Both are under continual development and have active on-line communities – an important consideration for developing expertise. Flight Control Hardware It is expected that flying in a swarm within a confined space will place more computational load on the flight control system than flying individually in the open air.  Additionally, further demands will need to be met for mapping and image recognition. Rather than place the full computational load on one ‘board’, flight stabilisation and control will be undertaken by one board and higher level functions such as mapping, image recognition and ‘where do I fly now?’ decisions will be the function of a second board.

  • For flight control, a Pixhawk board will be used (
    • This seems to have the most computational power of the open source hardware systems.
  • For higher level functions, a Rasbperry Pi will be used (
    • The existing Pi is well supported and a new Pi has recently become available with class leading computational power.

The two boards will communicate via a serial connection using the MAVLINK protocol ( Advantages

  • Improved modularity.  Better ability to update either component.
  • Reduced risk.  The flight control system will be (almost) entirely unaffected, reducing the possibility of errors being inadvertently introduced.
  • Reduced complexity.  Decoupling the functions simplifies the development task.

Next Steps Finalise the hardware/software combination and establish the MAVLINK communications protocol between the flight control system and the higher level decisions system (need a name for that!)