Lecture
This lecture discusses a nonmyopic planning approach under budget constraints, focusing on selecting hypotheses in ambiguous measurements scenarios. It covers the transition from unimodal Gaussian belief to multiple hypotheses, the use of a Gaussian mixture model, and the challenges of budget-constrained planning. The main contribution is an analytical framework providing guarantees on solution quality and performance acceleration, illustrated with a kidnapped robot scenario.