Page 33 - CBT 2018
P. 33
Routing algorithms in complex urban networks
Most urban transportation models do
not account for uncertainty in travel time when finding the shortest path from
an origin to a destination. In order to model the behavior of risk-averse users in networks with uncertain travel time,
a travel time reliability measure, such
as standard deviation of travel time σ, should be minimized along with mean travel time μ. In routing algorithms for vehicle navigation systems, inclusion of standard deviation makes the problem non-additive. Moreover, the level of risk- averseness may vary among users. Risk averseness can be modeled by some parameter λ representing the value of reliability; users more sensitive to travel time reliability would have a larger value of λ. CEGE researchers solved this problem by developing an algorithm to find the efficient frontier of the mean– standard deviation shortest path problem and determining the optimal reliable path for users with varying levels of risk-averseness. [48]
31
