Hi, My final idea is based on routing in road networks using reinforcement learning to provide the best optimised route from a starting point to a given destination. This is based on the *Shortest Route Problem. * The reinforcement agent is able to adapt in performing a high number of iterations with their dynamically changing environment. This would be beneficial when it comes to a dynamic road network which has updates on edge weights based on traffic data etc. This is not the same with path finding algorithms like dijkstra and A* algorithms which traverse the graph each time a request is made.
*Second Assumption* A novel graph preprocessing technique which could use time dependent contraction hierarchies with a customizable approach. This could help in storing edge weights for a relevant time frame as well as update edge weights based on abnormal weight changes due to blocks due to weather etc. Where only edge weights will be updated due to unexpected scenarios different from previous traffic data. Please help by providing your valuable input on where these steps would help in improving routing than using traditional algorithms and techniques. This would really help me for my final year project. Thank you in advance! Regards, *Anjula Paulus* *Informatics Institute of Technology* *Sri Lanka* *Mobile : +94 718136264/ +94 761514355 * *Email : anjula.2016...@iit.ac.lk <anjula.2016...@iit.ac.lk>*
_______________________________________________ OSRM-talk mailing list OSRM-talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/osrm-talk