Orchid Project

UAV Coordination Using MAX-SUM

To address the challenges related to the development of flexible autonomy, some initial work in collaboration with the Australian Centre for Field Robotics has been focused on the development and the deployment of decentralised coordination techniques on real Unmanned Aerial Vehicles (UAVs). The key idea was to develop a system in which first responders interacted in real time with a team of UAVs flying over the area of a disaster to request live aerial imagery of some specific sites. Each of these requests was modelled as a task. Each of these was characterised by a different importance incorporating the specifics of each site (e.g. a building on fire, a building about to collapse or a crowded area).

The challenge for the team of UAVs was then to coordinate in order to complete the highest number of highly important tasks. The importance of each task is determined by first responders, which are assumed to be provided with a Personal Digital Assistant (PDA) that they use in order to submit imagery requests. This problem was demonstrated in practice by running a set of flight tests involving two UAVs in three different scenarios:

  • Flight 1 – Homogeneous tasks: Two tasks are submitted at the same time, with the same properties, both the UAVs are aware of them. As shown in the video, in this setting each UAV goes to complete one task (represented as a whiteboard on the ground).
  • Flight 2 – Sequential Arrival of Tasks: Two tasks are submitted at different times. Initially, one low importance task is requested and after one minute a second task with a medium importance is submitted to the team. As shown in the second part of the video, the UAVs start going to the same task and revise their decision as soon as they both become aware of the second task.
  • Flight 3 – Heterogeneous Tasks: Two tasks are submitted at the same time, with the same properties. However, in this third test, one UAV is aware about only one of them. After one minute a new task is requested to both the UAVs. In this case, as depicted in the third part of the video, the UAVs go initially to one task each. Subsequently, as soon as they become aware of the third task, the UAV closer to it will be selected to complete it.

agents | citizen-science | applications | crowdsourcing | disaster response | smart grid | accountable information architecture |agent-based computing | agile teaming | disaster recovery |flexible autonomy | decentralised control | energy management |HACs | human-agent interaction |human computer interaction | human agent collectives | incentive engineering | machine learning |mechanism design | ORCHID |

Highlighted Publications

Browse Publications »

Disaster response

We are developing systems that allow first responders, unmanned ground and aerial vehicles, and software agents to work effectively together.

Learn more about Disaster response »

Smart Grid

We are developing novel algorithms and interfaces to optimise energy consumption and coordinate consumers and producers in the smart grid.

Learn more about Smart Grid »

Citizen Science

We are developing approaches that make full use of the skills, preferences and capabilities of citizen scientists.

Learn more about Citizen Science »