Artificial Intelligence for Operations Research

The Artificial Intelligence for Operations Research group (AI4OR) develops data-driven solutions for health care and autonomous systems in the domains of manufacturing, transportation, production and supply chain. Through close cooperation with leading industry and academic partners we excel at both the practical applications of operations research as well as strengthening the field’s theoretical basis through method development and testing.

The group’s research on autonomous systems is especially highlighted by its work in the operational phase of using automated guided vehicles (AGVs), mobile robots, satellites, and unmanned aerial vehicles (UAVs). In this area, we develop and implement planning, scheduling and routing algorithms that take stochastic elements and unpredicted events into account. Such capabilities are necessary to quickly plan and reschedule operations for improved execution. Achieving this requires innovative methods and tools that can find robust solutions online, offline and hybrid mode. This is truly a challenging topic that requires novel approaches.

For the medical domain, the group develops prediction models, and clinical decision support tools for patients with e.g. cancer (Febrile neutropenia, Lymphoma, etc.) and congenital red blood cell disorders (hemoglobinopathies). The decision support tools are created by using artificial intelligence/machine learning methods to predict risk scores for different individuals to classify patients with hemoglobinopathies or divide into low- and high-risk cancer groups.

The group excels in understanding the decision environment of complex industrial problems and developing appropriate solution approaches (based on e.g. genetic algorithms, heuristics, exact approaches, and constraint programming techniques) implemented as prototypes and decision support systems. This is highlighted by the group’s work on:

1. earth observation satellite scheduling for improved climate monitoring and security operations
2. maritime search and rescue planning with drones that improve the probability of survival,
3. addressing problems with ambulances location and post–mass-casualty deployment by proposing an algorithm, that incorporates realistic demand scenarios, and optimizing patient prioritization and hospital assignment by column generation techniques,
4. proposing improved policies for large cargo-vessels and container terminals to reduce both fuel and emissions,
5. analyze of supply chains with regards to the green transition,
6. and maintenance prediction for wind turbines to reduce costs and increase reliability for large-scale wind energy operations.