Verifiable Safety for Autonomous Vehicles
Verifiable safety in Autonomous Vehicles (AV) remains an elusive goal. One major reason for this is the reliance on unverifiable neural networks to perform safety critical tasks. We propose Synergistic Redundancy architecture, which decouples safety- and mission-critical tasks, while leveraging synergies between the partially redundant safety and mission subsystems.
Publications
Bansal, A., Yu, S., Kim, H., Li, B., Hovakimyan, N., Caccamo, M. and Sha, L., 2022. Synergistic Redundancy: Towards Verifiable Safety for Autonomous Vehicles. arXiv preprint arXiv:2209.01710. Under peer review.
Bansal, A., Kim, H., Yu, S., Li, B., Hovakimyan, N., Caccamo, M. and Sha, L., 2022. Verifiable Obstacle Detection. arXiv preprint arXiv:2208.14403. Accepted for publication at the 33rd IEEE International Symposium on Software Reliability Engineering (ISSRE) 2022.
Bansal, A., Singh, J., Verucchi, M., Caccamo, M. and Sha, L., 2021, June. Risk Ranked Recall: Collision Safety Metric for Object Detection Systems in Autonomous Vehicles. In 2021 10th Mediterranean Conference on Embedded Computing (MECO) (pp. 1-4). IEEE.