Category: Conference

Zhenjiang Mao presented his work on learning-enabled safety prediction (poster, paper) at the 6th Annual Conference on Learning for Decision and Control (L4DC 2024) in Oxford, UK. Reportedly, the attendees like math more than he does.  Citation:  Zhenjiang Mao, Carson Sobolewski, Ivan Ruchkin. How Safe Am I Given What I See? Calibrated Prediction of Safety…

The papers were on foundation world models and language-enhanced OOD. The audience response was, reportedly, positive and encouraged the implementation on physical robotic systems. Citations: 

In the first presentation of ICCPS 2024, Ivan showcased a method to repair a neural network controller while preserving its verification results.  Citation:  Pengyuan Lu, Matthew Cleaveland, Oleg Sokolsky, Insup Lee, Ivan Ruchkin. Repairing Learning-Enabled Controllers While Preserving What Works [Arxiv] [Github] [Slides]. In Proceedings of the International Conference on Cyber-Physical Systems (ICCPS), Hong Kong, China, 2024. 

Update 1: an extended version of this talk was given at a UF MAE Affiliate Seminar. The recording can be found here (UF login required). Update 2: another version of this walk was given at the ETH Autonomy Talks (video).  Update 3: yet another version of this talks was given as a CNEL Seminar. The…

Shown above is a 5-step workflow of our causal repair: (1) Extract the behaviors of a learning component as an I/O table. (2) Encode the dependency of the desired property outcome on the I/O behaviors with a Halpern-Pearl model. (3) Search for a counterfactual model value assignment, revealing an actual cause and a repair. (4)…

Shown above is our conservative monitoring approach that leverages probabilistic reachability offline and combines it with calibrated state estimation. Citation: Matthew Cleaveland, Oleg Sokolsky, Insup Lee, Ivan Ruchkin. Conservative Safety Monitors of Stochastic Dynamical Systems [ArXiv] [Springer] [Slides]. In Proceedings of the NASA Formal Methods Symposium (NFM), Houston, TX, 2023.