GTCC Autobot
| Reference number | |
| Coordinator | Blue Mobile Systems AB |
| Funding from Vinnova | SEK 500 000 |
| Project duration | October 2019 - July 2020 |
| Status | Completed |
| Venture | AI - Competence, ability and application |
| Call | Start your AI journey! |
Important results from the project
The aim of the project was to apply and evaluate ML and AI methods to improve the response time when sending alarms from alarm centers to security guards in the field. The goal was to implement these methods for our product GuardTools Command & Control. The results from our test runs were not significant enough to proceed with implementation.
Expected long term effects
The general conclusion from the project is that it is very difficult to train reinforcement learning agents to be significantly better than the naïve policy, especially if the goal you have is to minimize the average response time. However, in some circumstances we manage to produce agents that are slightly better than the naive policy.
Approach and implementation
The project was divided into two main parts, AP1 and AP2. AP1 included data processing and structuring, AP2 focused on method development. BMS has for many years collected large amounts of alarm data. AP1 ensures that the collected data is made available and structured so that the FCC can develop and validate the ML and AI methods that will be analyzed within the project. The goal of AP2 is to evaluate and develop ML and AI methods to make data-driven decisions about how resources should be distributed when different types of alarms arrive at alarm centres.