Abstract :
Collision Avoidance for General Aviation Thomas Billingsley, Lincoln Laboratory, MIT • Model CA as Markov Decision Process (MDP) • Encode performance constraints of GA • Optimize for total overall expected cost: — Near mid-air collision, alert, resolution reversal Decomposition Method for Optimized CA with Multiple Threats James Chryssanthacopoulos, Lincoln Laboratory, MIT • Pairwise vs. global • Extend MDP to pairwise decomposition — Command arbitration: optimal pairwise action — Utility fusion: optimal pairwise utility (e.g. max-sum)
