Formal Verification of Neural Network Controllers for Collision-Free Flight

Genin, Daniel; Papusha, Ivan; Brulé, Joshua; Young, Tyler; Mullins, Galen E.; Kouskoulas, Yanni; Wu, Rosa; Schmidt, Aurora

doi:10.1007/978-3-030-95561-8_9

Cited by 2 publications

(1 citation statement)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, we are pursuing methods for constraining the learned policies of safe neural network controllers using the safety predicates from this work, such as in [7]. The safety predicates developed here may be used to guide the training of such networks, verify the correctness of the network policies, and could someday be used directly in the optimization of such neural controllers.…”

Section: Future Workmentioning

confidence: 99%

Envelopes and Waves: Safe Multivehicle Collision Avoidance for Horizontal Non-deterministic Turns

Kouskoulas,

Machado,

Genin

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

We present an approach to analyzing the safety of asynchronous, independent, non-deterministic, turn-to-bearing horizontal maneuvers for two vehicles. Future turn rates, final bearings, and continuously varying ground speeds throughout the encounter are unknown but restricted to known ranges. We develop a library of formal proofs about turning kinematics, and apply the library to create a formally verified timing computation. Additionally, we create a technique that evaluates future collision possibilities that is based on waves of position possibilities and relies on the timing computation. The result either determines that the encounter will be collision-free, or computes a safe overapproximation for when and where collisions may occur. IntroductionAutonomous and semi-autonomous systems that control ground vehicles, boats, and aircraft all need to rea-This research was partially funded under the sponsorship of the Federal Aviation Administration Traffic Alert & Collision Avoidance System (TCAS) Program Office (PO) AJM-42 under contract number DTFAWA-11-C-00074 as well as internal funds from the

show abstract

Section: Future Workmentioning

confidence: 99%