sminT

sminT is a planning framework in computer science and operations research for minimizing expected completion time in the presence of uncertainty. The name stands for stochastic minimum time and refers to problems where traversal or transition times are random variables with known distributions. sminT formulations typically seek a policy or path that minimizes the expected time to reach a designated goal from a start state, optionally under reliability constraints or time budgets.

History and scope: The concept emerged from studies in stochastic path planning and robust optimization in the 2010s. Early work focused on discrete-time, discrete-space networks; later research extended sminT to continuous spaces and multi-agent scenarios, and to risk-sensitive objectives.

Modeling and methods: A common model encodes each state as a location and a time, and actions produce stochastic transition times. Solution methods include dynamic programming techniques such as value or policy iteration, often with discretization of time and convolution of time distributions. For large-scale problems, approximate and sampling-based methods—Monte Carlo simulation, scenario trees, or heuristic-guided search—are used to improve scalability. Variants incorporate constraints on maximum allowable delay, risk measures, or time windows.

Applications: sminT has been applied to transportation networks, logistics and supply chains, autonomous robotics, and emergency response planning, where uncertain travel times can significantly affect performance.

Limitations: The approach relies on accurate probabilistic models of travel times; computational requirements grow with network size and distribution complexity; real-time updates may require rapid re-planning.

See also: stochastic shortest path, Markov decision process, dynamic programming, robust optimization.