Research

<Towards More Autonomy for Unmanned Autonomous Vehicles>

The focus of this research lab is on introducing more autonomy into unmanned vehicles by developing autonomous decision making (mission planning/task allocation), guidance (path planning) and control algorithms. 

Fig. 1. Required functions for unmanned autonomous vehicle operation and the concept of DARPA’s heterogeneous aerial reconnaissance team.

It is anticipated that unmanned vehicles will be widely used within military and civilian operations and have a profound influence in our daily life. A group of small unmanned vehicles are of special interest due to their versatility, flexibility and ability to co-operate to achieve common goals with inherent redundancy. Before fully realising the potential that unmanned vehicles bring, these systems should be able to achieve a similar level of autonomy and safety as manned systems in order to make them accepted by public and regulatory authorities. For instance, currently, most of drone operations are managed by a human pilot with limited autonomy; thus, they require a significant human workload and cost. In addition, compared with a human pilot residing in the vehicle, a major safety concern of unmanned vehicles is the inevitable reduction in situational awareness (i.e. understanding of vehicle’ states and the surrounding environment) of the operator remotely located in a control station.

In this context, to implement a fully autonomous cooperative unmanned vehicle system in a safe, timely and appropriate manner with a little or no human intervention, two aspects of unmanned vehicles need to be addressed: i) increasing the level of autonomy with high-level mission and path planning algorithm developments; and ii) improving situational awareness with sensor/information fusion techniques with domain knowledge for safer operations and better mission performance, as illustrated in Fig. 1. Although vehicle dynamics and control is also an important area, our focus is on high-level planning and signal processing as they are still immature yet critical for the autonomous operation of unmanned vehicles. This research is expected to help regulatory authorities to understand the behaviour of unmanned vehicles and the risks/safety issues caused by increasing the level of autonomy. The situational awareness will help end users and operators to determine proper levels of autonomy in response to the change of real operation scenarios.

One of representative real world applications that require the autonomous cooperative unmanned system would be the intelligence, surveillance, and reconnaissance (ISR) mission, as illustrated in Fig. 1 (right). To perform the ISR mission with drones autonomously, various algorithms need to be developed/integrated including: resource/task assignment, search, target detection, information analysis, target tracking, and team communication/networking. The autonomous cooperative unmanned system is not only applicable to ISR but also to numerous civil missions (e.g., search and rescue, environmental monitoring, infrastructure protection, police law enforcement, etc.).

Here are some examples of the laboratory's research work:

Decision Making, Guidance (Path Planning) and Control
  • Area and road-network search

  • Autonomous search and source term estimation of hazardous material using a mobile sensor platform

  • Persistent moving ground target following

Fig. 2. Target tracking guidance (path planning) using the vector field approach for multiple drones considering the limited sensor field-of-view (FOV) and kinematic constraints. 



  • Airborne communication relay using UAVs
Fig. 3. Communication-aware ground vehicle tracking which maintains communication between a moving ground convoy and a ground control station (GCS) using geometric visibility and machine learning (Gaussian process).
 
Fig. 4. Optimal dynamic positioning of drones to maximise the communication performance via numerical simulation and indoor flight test.




Estimations and Sensor/Information Fusion Based on Bayesian Framework
  • Autonomous airborne surveillance of ground traffic behaviour
Fig. 5. Realistic traffic simulation software (S-Paramics) with UK village map used in the project and a decision logic to classify the temporal trajectory of a ground car into driving behaviour.
Fig. 6. Airborne behaviour monitoring process to identify suspicious cars in the ground traffic.

  • Moving target (e.g. ground cars or ballistic missiles) tracking

Other Related Research on Unmanned Air and Ground Vehicles
  • Bio-inspired swarming/flocking algorithms


Fig. 7. Illustration on the EC FP7 Swarm-Organ project and micro robots, called Kilobots, (the size is about 3cm) used in the project. 

  • Development for vision-based autonomous flight of insect-mimicking subminiature drone