PAC-Bayes control for obstacle avoidance with Parrot SWING

Results from: "PAC-Bayes Control: Learning Policies that Provably Generalize to Novel Environments" (Authors: Anirudha Majumdar, Alec Farid, and Anoopkumar Sonar).
Link to paper:

This video shows an obstacle avoidance policy designed for a Parrot SWING drone using PAC-Bayes theory. The policy comes with a guaranteed expected success rate on novel environments (approximately 88.4% for this setting). The empirical success rate is approximately 90% (18/20 trials).

SWING