MPC-generated approximate value labels guide a DeepReach-style network to learn Hamilton-Jacobi reachability solutions, yielding larger verified safe sets in 2D, 7D, 13D, and 40D systems.
Algorithm for overcoming the curse of dimension- ality for time-dependent non-convex Hamilton–Jacobi equations arising from optimal control and differential games problems
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Bridging Model Predictive Control and Deep Learning for Scalable Reachability Analysis
MPC-generated approximate value labels guide a DeepReach-style network to learn Hamilton-Jacobi reachability solutions, yielding larger verified safe sets in 2D, 7D, 13D, and 40D systems.