Ford Safety Insight Platform (FSIP) promises to vastly reduce the time and cost for cities to conduct meaningful analysis of their transportation data - resulting in savings that could be put towards actionable improvements for traffic safety.
Data on traffic accidents, vehicle volumes and other traffic events collected by the city can be layered on top of Ford’s own proprietary connected vehicle data from the FSIP. These layers of data are fed into the system, which can then 1) visualize the incidents – for example repetitive crashes broken down by related factors, such as alcohol, speed, cyclists, pedestrians; 2) identify hotspots where accidents most commonly occur; and 3) simulate solutions.
Windsor city decision makers can then use the results from FSIP to investigate costs and benefits with certain infrastructural changes, paving the way for data-driven decisions that could improve safety to a greater extent for the same amount of investment.
FSIP seems to be a promising tool and I am especially interested in learning about the validation of its method. Identification of causes and simulations of solutions sounds very advanced when considering the complexity of traffic accidents, which includes factors such as infrastructure, vehicle safety sophistication, traffic rules, traffic norms, and last but not least human cognition.
The success of an AI tool such as FSIP is largely dependent on how much and what data has been used to train the algorithms. In other words, the identification of causes and the producing of solutions are limited to the data that has been fed into the algorithms of FSIP. This is not to say that establishing viable causal links in traffic accidents cannot be done on more general levels, but it would require impressive amounts of training datasets to produce sufficiently substantiated results with the help of machine learning algorithms.