Motion planning with applications to air traffic management
Date of Issue2017-12-13
School of Mechanical and Aerospace Engineering
Robotics Research Centre
Air Traffic Management Research Institute, NTU
This report introduces a new algorithm to incorporate robot temporal goal specifications in real-time air traffic management. The approach is inspired by current robot motion planning theories including Probabilistic Roadmap and Rapidly-exploring Random Tree, as well as the cutting-edge model checking methods such as product automaton verifications. The core component of the algorithm lies in the temporal planner. It allows the air traffic controller to pre-specify a set of intermediate goals or milestones for both single and multiple aircrafts. These specified commands expressed in standard Linear Temporal Logic(LTL) formula will then be translated into a standard automaton for model checking purposes. Subsequently, aircraft trajectories can be generated to fulfil pre-specified missions accordingly based on the shortest path search results in the product automaton map. In particular, demonstrations show that a 5-aircraft decoupled planning problem with different individual specifications can be resolved within 0.9 seconds, while 1-aircraft decoupled planning problem can be resolved within 0.3 seconds. The coupled planner is also successful in resolving the joint planning for 2 aircrafts with a composite LTL formula specification. The proposed algorithm can be used for all series of motions as long as the specification is in standard LTL format. It helps to address planning issues especially with a time sequence of multiple goals. As long as a solution exists, our approach guarantees to generate a corresponding optimal path which satisfies the prescribed LTL formula for both individual and multiple robot planning cases.
Final Year Project (FYP)
Nanyang Technological University