Jendrik Seipp, WASP researcher and senior associate professor at Linköping University, has been awarded SEK 15 million from Stiftelsen för Strategisk Forskning (SFF) to develop an AI planning system that fully leverages multi-core, parallel computation.
“I’m very pleased to have been given this opportunity to pursue my vision. I believe that this is absolutely crucial for the future of AI planning and its applications,” says Jendrik Seipp, senior associate professor at the Department of Computer and Information Science.
Potential applications range from logistics and power distribution to cybersecurity. These are societal functions that require advanced planning to operate in the most resource-efficient way possible. Artificial intelligence can be a powerful tool for improving such planning, but one major problem remains to be solved.
“Today’s most advanced AI planning systems can only use one core of a computer’s processor. This means that they handle one task at a time, one after the other. This limits their ability to scale up the system,” says Jendrik Seipp.
Solving complex problems
To solve this problem, he has been granted SEK 15 million from the Swedish Foundation for Strategic Research (SSF) as part of an initiative to support future research leaders. The selected researchers are to conduct cutting-edge research as well as possess leadership skills and demonstrate a willingness to ensure their research can be used outside academia. Out of 213 applications, 16 researchers were granted funding.
According to Jendrik Seipp, the biggest challenge is designing data structures that work with parallel solutions and can be scaled up without being slowed down by workload imbalance between processor cores, synchronisation issues, or memory limitations.
“If we succeed, it will be possible to solve far more complex problems than today’s sequential systems can manage. We want to establish an open, broad framework for parallel planning that will become the standard foundation for research and industry,” says Jendrik Seipp.
The long-term goal is to achieve reliable AI planning that is understandable and explainable and also has some kind of human involvement. In the future, this could mean faster and more environmentally friendly home deliveries through efficient route planning, safer autonomy in robots and vehicles via real-time planning, and cheaper, cleaner energy use through smart scheduling of consumption and production.
Text and image courtesy: Linköping University.
Text: Anders Törneholm
Published: October 29th, 2025
[addtoany]