The paper proposes decentralized deconfliction algorithms deployed on multiple autonomous aerial vehicles in freeflight operations. The paper provides two separate algorithms for collision avoidance - one based on the iterative peer-to-peer negotiation solving a singular collision and second based on multi-party negotiation about a cluster of collisions. The presented decentralized algorithms allow the vehicles operating in the same area to utilize the given airspace more efficiently. The algorithms have been developed and tested on a multi-agent prototype and the properties of both algorithms are discussed on a set of large scale experiments. The authors use the agent based approach to implement the cooperative collision avoidance by a local collision avoidance implementation. The detailed specification of the airspace domain is stated in the section 2. The restriction to use only decentralized algorithm is given by the fact that airplanes in the domain can communicate only with planes located nearby. Well known techniques for collision avoidance based on the potential fields are not suitable for the domain (3D, no-flight zones, the need smooth flight plan also in its first derivation, allowed speed changes) due to their complexity and because of the complicated dynamic mission airplane specification. The main contribution of the paper is in the novel multi-party collision avoidance (MPCA) algorithm (section 6) for the described airspace domain. The algorithm removes the iterations from the iterative peer-to-peer (IPPCA) algorithm (new formalized description is in the section 5) for multicollisions among several airplanes. The use of searching for the best combination of evasion applications used in the proposed algorithm leads to better solutions (in means of used criterion) and thus for better utilization of given airspace than in the IPPCA. The 'See' ability of the algorithms is the same, see the section 3. Both algorithms use the same set of possible evasion manoeuvres (section 4). The properties of the proposed MPCA are studied in experimental way by a comparison to the existing IPPCA (section 7). The IPPCA algorithm was selected for the comparison because it provides the best result from all implemented ones in the domain multi-agent simulator.
Decentralized algorithms for collision avoidance in airspace
Dezentrale Algorithmen zur Kollisionsvermeidung im Luftraum
2009
8 Seiten, 4 Bilder, 13 Quellen
Conference paper
English
Hierarchical, Hybrid Framework for Collision Avoidance Algorithms in the National Airspace
British Library Conference Proceedings | 2008
|