Data-Driven Optimization of Automated Speed Enforcement Logistics

Abstract

Canada’s collision fatalities are about 2000 lives a year, decreasing in the last decade, reaching 1745 in 2020 due to initiatives like Vision Zero. Among municipalities’ priorities is to enforce speed limits to reduce speeding-induced traffic fatalities, constituting 27 % of all traffic fatalities in Canada. An emerging strategy toward this goal is the deployment of , which detects violators through speed cameras positioned alongside designated roads. Empirical evidence from existing Automated Speed Enforcement (ASE) practices shows that the number of citations drops each month as driver become aware of camera locations and lower their driving speeds. Hence, ASE cameras are often relocated in cycles to expand their reach to more places and further deter speeding violations. The complexities of deployment lie in choosing camera locations and cycle duration, which have the highest deterrence impact on speeding during a planning period. This study proposes a data-driven model to classify camera site locations based on the effectiveness of ASE enforcement. Then, a Markov decision process optimization model is presented to find the optimal camera locations at each cycle and the length of the cycles for minimizing speed violations across the entire transportation network, considering limitations such as the number of available cameras.