Verifying Aircraft Collision Avoidance Neural Networks Through Linear Approximations of Safe Regions
This work addresses the certification challenge for neural network compression of aircraft collision avoidance tables, which is critical for safety-critical aviation systems.
The authors used linear approximations of safe regions to verify that neural network-based collision avoidance systems do not issue unsafe advisories, and found thousands of unsafe counterexamples in a notional policy.
The next generation of aircraft collision avoidance systems frame the problem as a Markov decision process and use dynamic programming to optimize the alerting logic. The resulting system uses a large lookup table to determine advisories given to pilots, but these tables can grow very large. To enable the system to operate on limited hardware, prior work investigated compressing the table using a deep neural network. However, ensuring that the neural network reliably issues safe advisories is important for certification. This work defines linearized regions where each advisory can be safely provided, allowing Reluplex, a neural network verification tool, to check if unsafe advisories are ever issued. A notional collision avoidance policy is generated and used to train a neural network representation. The neural networks are checked for unsafe advisories, resulting in the discovery of thousands of unsafe counterexamples.