The BBC is reporting that the rail industry is investigating the use of satellite navigation to provide dynamic traffic management. While proponents of this technology are pushing the efficiency and safety aspects, I'm interested in the increased autonomy and intelligence this gives to the trains themselves.
Could a decentralised control system be more efficent than the current top down model? An agent architecture, where individual trains negotiated their way across the network in a collaborative agent model would be "interesting" to say the least, and perhaps not that far fetched. Similar architectures have been used to move packets of data across fairly complex network topolgies, and after all, conceptually there isn't that much difference between moving a few hundred tons of train and a few hundred bytes of data...
Of course, what a person travelling on the railways is interested in is a single train, we don't usually care about the network as a whole. Perhaps our selfish nature could work in our favour for once as your personal agent, running on your 3G mobile phone or PDA, could add it's vote to the agent on the train and make your train more important, improving its journey across the network. In the peer-to-peer negotiations for priority, a train with many people waiting for it would get right of way.
More prosaically, the exact location of the train you're waiting for would be known to your agent and would be part of your augmented reality.
No comments:
Post a Comment