Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore reachability analysis for complex control systems in this 13-minute conference talk. Learn about the speaker's work on OVERT.jl for approximating nonlinear functions with piecewise linear functions in JuMP.jl, and OVERTVerify.jl for computing reachable set hulls in nonlinear systems with neural network control policies. Discover the concept of temporal refinement in reachability analysis, which balances precise symbolic queries and imprecise concrete queries within a fixed time budget. Understand how this approach leverages JuMP.jl's early stopping interface and solution gap checks to manage computational resources effectively. Gain insights into a software tool that computes reachable sets for complex systems, including those with neural networks, adapting fidelity based on available computational budget.
