The Cooperative Sorting Strategy for Connected and Automated Vehicle Platoons - Online Processing A* Algorithm
Typ
Examensarbete för masterexamen
Program
Publicerad
2020
Författare
Kong, Xiangyu
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
Along with the fast development of urbanization, the optimization of transportation
plays an increasingly important role for solving traffic jam problem. Based
on the assumption of fully connected and automated vehicles(CAVs), the optimal
control and design of CAVs is one of the most effective ways for improving traffic
efficiency as sorted CAVs platoons could help optimise the capacity of intersections.
This research focus on optimizing the sequential actions of vehicle platoons for this
given aim state. With the help of the perception and localization, multi-line vehicle
platoons are abstracted into the permutation. The optimization object is the length
of the trajectory from start permutation to the target permutation. A* algorithm
wrote in Python could perform better with the help of the hash table and PyPy3.
However, the time complexity problem still exists, which blocks it from extensive
applications. Model-free Reinforcement Learning (RL) method suffers from the large searching
space and low sampling efficiency a lot and Mento Carlo Tree Search (MCTS) lacks
suitable criteria. In this essay, online processing A* (OPA*) algorithm is original
promoted for solving time complexity problem with the sacrification of the optimality.
The comparison and evaluation of OPA* and A* methods mainly focus on the
performance and time consumption. OPA* could give stable and scalable results
which make it possible for industrial usage.
Beskrivning
Ämne/nyckelord
sorting , hash table , heuristic algorithm , reinforcement learning , online processing A*