Get Started
Research questionHow can safety-critical controllers update differentiable boundaries as data-defined unsafe regions evolve?Safety-critical controllers need differentiable boundaries to distinguish allowable states from unsafe ones. When unsafe regions are represented by changing data, the boundary must adapt without losing conservative coverage or safety guarantees.
Alignment & Safety
Machine Learning
Robotics
Latest papersRecent research connected to this question, newest first.Learning-based Adaptive Safety-Critical Control With Evolving Unsafe RegionsApplies to control-barrier-function-based safety control, including incremental changes to unsafe-region samples. The evidence includes analytical regularity and safety conditions and a two-link manipulator experiment involving collision-free target reaching; broader deployment settings are not established.research paper · Sep 3, 2026
Related questions
How can safety-critical control prioritize multiple uncertain risks while certifying how far a policy is from optimal?How can vision-language models correct unsafe generations token by token without disrupting safe reasoning?How can we compute the largest set of states admitting a safe policy to reach and stay in a target despite bounded disturbances?How can safety-critical AI demonstrate complete coverage of high-dimensional operational design domains?
Home
Topics
Search
Library