Registered Data

[02374] Embarrassingly-parallel optimization algorithms for high-dimensional optimal control

  • Session Time & Room : 2E (Aug.22, 17:40-19:20) @F401
  • Type : Contributed Talk
  • Abstract : Developing efficient algorithms for Hamilton--Jacobi partial differential equations $(\text{HJ PDEs})$ is crucial for solving high-dimensional optimal control problems in real time but notoriously tricky due to the so-called curse of dimensionality. In this talk, we present novel grid-free and embarrassingly-parallel optimization algorithms for solving a broad class of HJ PDEs relevant to high-dimensional state-dependent optimal control problems. We illustrate their performance and efficiency on large-scale multi-agent path planning problems.
  • Classification : 49L12, 65K10, 90C30, 49M29, 49M37
  • Format : Talk at Waseda University
  • Author(s) :
    • Gabriel Provencher Langlois (New York University)
    • Jerome Darbon (Brown University)