Previous research [1][2][3][4][5][6][7][8][9] has developed methods for the optimization of routing based on the criteria of travel time and/or on the cost of travel and/or the distance traveled. Routing can be in various modalities, such as by car, on foot, by bicycle, via public transit, or by boat. A typical method of routing involves building a graph comprised of street segments, assigning a normalized weighted value to each segment, and then applying the weighted-shorted path algorithm to the graph in order to find the best route.
Routing can take into account preference parameters in addition to time and distance. For example, routing suggestions can include c consideration pertaining to the reduction of the risk of encountering violent crime. For example, a user desires a leisure walk via a safe route from her hotel in an unknown city. Here we present a method to quantify such user preferences and the risks of encountering crime and to augment the standard routing methods by giving weight to said safety considerations.
Galburn et al. [4] have utilized crime data to optimize the safety aspect of navigation within a city. Their case study involved urban crime data from Illinois and Pennsylvania. Their proposed risk model for the street network within a city facilitated estimating probabilities of criminal incidents that the traveler may encounter on any road segment. In their approach, the same importance is assigned to the path traversal time and the crime incident risk. Their method solves a dual-objective shortest-path problem.
Here we presented an improved method to cooptimize crime avoidance with other criteria. The proposed method's advantages, in comparison to other crime-avoidance routing algorithms, include weighing crime types with respect to their potential detrimental value to the user, with temporal qualification and quantification of crime and its statistical aggregation at the geographic resolution down to a city block.
The following figure shows traditional routing optimizing the time and/or distance. Here we present an improved method to cooptimize crime avoidance with other criteria. The proposed method's advantages, in comparison to Galburn [4] and the other crime-avoidance routing algorithms, include: (1) weighing crime types with respect to their potential detrimental value to the user, (2) with temporal qualification, (3) quantification of crime and its statistical aggregation at the geographic resolution down to a city block, and (4) evaluation of the crime detriment to the user in each segment by considering the needs, exposure, and preferences of the user rather than merely considering the general crime incidence statistics. For example, violent crime committed outdoors have a higher impact, and severe violence, such as homicide in the street, have the highest impact. Crimes without a direct unrelated victim, such as code violations or embezzlement, have no impact on travelers. Pick-pockets have an impact on travelers in walking mode but minimal impact on travelers by car. Non-statutory rape may be of high concern to a woman walking alone. For each type of traveler and travel modality, the present method provides default formulas for the evaluation of crime detriment in each segment. Additionally, the user may modify the formula by assigning greater or lesser importance to various types of crimes.
In order to quantify crime risks for each street segment, we count police reports that occurred close to that segment during a set period of time, e.g., a particular year of reference, counting only violent and property crimes of the type that would directly affect the traveler (e.g., exclude domestic violence, exclude insider trading, exclude code violations, exclude statutory rape) and can further assign weights to various crime crimes based on the impact it may have in the traveler. The following is an example of a query to a crime database for an area in mid-Miami Beach. The following is a tabular output of the query:
The importance of querying for only specific types of crime (and weighting them) is demonstrated by the following query, whose results are mostly crimes that have no bearing on the prospective traveler. Applying said prior-art method to the herein proposed weighting selection problem, three objectives (A=time, B=cost of travel, and C=crime avoidance) are presented in a triangular fashion on a touch screen. Sub-figure 1 shows the underlying principle of the establishment of a single weight wA for Objective A; Sub-figure 2 combines three objectives into a single triangle, allowing for the establishment of a tri-variable weight function (wA, wB, wC). By applying a finger gesture, the user moves an indicator freely inside the triangle (see Sub-figure 3). The position of the indicator establishes a tri-variable weight function, which in further steps, is then used as input for a co-optimization algorithm. When the user is satisfied with the established weights, she indicates this, e.g., by pressing a touch screen button labeled "Go."