Abstract
The idea of a Cloned Controller to approximate optimised control algorithms in a real-time environment is introduced. A Cloned Controller is demonstrated using Linguistic Decision Trees (LDTs) to clone a Model Predictive Controller (MPC) based on Mixed Integer Linear Programming (MILP) for Unmanned Aerial Vehicle (UAV) path planning through a hostile environment. Modifications to the LDT algorithm are proposed to account for attributes with circular domains, such as bearings, and discontinuous output functions. The cloned controller is shown to produce near optimal paths whilst significantly reducing the decision period. Further investigation shows that the cloned controller generalises to the multi-obstacle case although this can lead to situations far outside of the training dataset and consequently result in decisions with a high level of uncertainty. A modification to the algorithm to improve the performance in regions of high uncertainty is proposed and shown to further enhance generalisation. The resulting controller combines the high performance of MPC–MILP with the rapid response of an LDT while providing a degree of transparency/interpretability of the decision making.
Highlights
Behavioural cloning [1,2,3] has been used to imitate human control of systems that are difficult to model analytically
An alternative application is in complex real-time systems where the control policy must be determined very rapidly, whilst it may be possible to model such systems, deriving the control policy in real-time is often challenging such as the control of obstacle avoidance for Unmanned Aerial Vehicles (UAVs) [4,5]
We present novel extensions to the fundamental LID3 algorithm to account for circular attribute domains and to predict discontinuous output functions that are inherent to the UAV path planning problem
Summary
Behavioural cloning [1,2,3] has been used to imitate human control of systems that are difficult to model analytically. The remainder of this paper is organized as follows: Section 2 introduces the relevant background of the applied methods and formally defines the problem statement; Section 3 presents the MPC reference controller used to generate the training data; Section 4 presents several modifications to the standard LID3 algorithm necessary for successful navigation in scenarios containing one or two hostile regions; Section 5 demonstrates successful generalisation to scenarios containing a larger number hostile regions and investigates factors that influence the performance of the cloned controller; Section 6 demonstrates how the semantic nature of the algorithm promotes good interpretability and presents some metrics to measures the interpretability of the proposed controller; Section 7 presents an extension to the algorithm to improve performance in regions of high uncertainty and compares the computational complexity of the new algorithm with that of the optimisation; Section 8 highlights the main findings and conclusions of the paper
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.