In June, the International Planning Competition held every two years will take place in New Hampshire. It is divided into four categories, the most significant of which is considered the “deterministic track”. This is for programs designed to eliminate any element of chance from automated planning in a wide range of fields, such as logistics, robot manipulation, satellite movement and transport.
Yes, the big money in AI is not about running missiles or playing Tron, it is about getting your UPS shipments more efficiently. For 2014, there have been some 75 entrants to the deterministic track. They have submitted their problem-solving programs which will be tested on tasks and application domains that were devised and adapted by a University of Huddersfield trio - Dr. Mauro Vallati, a two-time winner of the Learning Track of the contest, Dr. Lukas Chrpa and Professor Lee McCluskey. Vallati devised a domain concerned with the scheduling of aircraft movement at an airport. McCluskey has been involved in a pan-European research project named Autonomic Road Transport Support Systems and for the International Planning Competition. This has resulted in a road transport problem for entrants’ problem-solving programs to solve.
Every domain within the deterministic track contains up to 300 problems that have to be addressed by the software and as a result it has to be run for up 150 hours when it is tested. This process has to be repeated for all of the 75 entrants. They estimate that the results and the names of the winners will be known by the end of May.
“Competitions are very good things especially in our academic area. You get some highly motivated people spending a lot of time trying to win and this pushes the technology forward,” said McCluskey.