A spatiotemporal risk prediction of wildlife-vehicle collisions using machine learning for dynamic warnings

INTRODUCTION

The technology in the automotive industry is becoming increasingly safer in the age of automated driving, but the number of accidents is still high, especially in wildlife-vehicle collisions (WVCs). To better avoid these accidents, patterns involved in these accidents must be detected.

METHOD

This paper presents a spatiotemporal risk prediction of WVCs, including various road and environmental characteristics. A process of data preparation using GIS automated by Python scripts was developed to enable a spatiotemporal link of diverse features for the subsequent predictive data analysis. Different machine learning (ML) approaches were applied- random forest (RF), feedforward neural networks (FNN), and support vector machine classifier (SVM) - including automated ML to predict the risk of WVCs. Therefore, a dataset of approximately 731,000 accidents reported to the police in Bavaria over a period of 10 years (2010-2019) was used. In addition, non-accidents were randomly generated in Python over time and space for the supervised ML processes. As the actual risk probability for WVCs and non-WVCs is not entirely known, the impact of different training ratios between accidents and non-accidents was tested on the risk prediction quality (RPQ) (25%, 50%, 75%, 90% WVCs) of the double-weighted sensitivity and single-weighted specificity rate.

RESULTS

The test yielded high mean values of RPQ as an indicator for a suitable WVC prediction. Both RF (86.6%) and FNN (86.7%) were identified as suitable choices for WVC risk prediction in terms of RPQ. The SVM yielded a lower prediction quality, even though acceptable results could be achieved within a shorter runtime.

PRACTICAL APPLICATIONS

Spatial transferability was verified since the algorithm was trained on the dataset of Bavaria (excluding Upper Bavaria) and successfully tested in Upper Bavaria. WVC forecasts were also proven through training with datasets from 2010-2017 and in prediction for 2018 and 2019.

Impact of Road Fencing on Ungulate–Vehicle Collisions and Hotspot Patterns

The number of road traffic accidents decreased in Lithuania from 2002 to 2017, while the ungulate–vehicle collision (UVC) number increased and accounted for approximately 69% of all wildlife–vehicle collisions (WVC) in the country. Understanding the relationship between UVCs, traffic intensity, and implemented mitigation measures is important for the assessment of UVC mitigation measure efficiency. We assessed the effect of annual average daily traffic (AADT) and wildlife fencing on UVCs using regression analysis of changes in annual UVCs and UVC hotspots on different categories of roads. At the highest rates, annual UVC numbers and UVC hotspots increased on lower category (national and regional) roads, forming a denser network. Lower rates of UVC increase occurred on higher category (main) roads, forming sparser road networks and characterized by the highest AADT. Before 2011, both UVC occurrence and fenced road sections were most common on higher-category roads. However, as of 2011, the majority of UVCs occurred on lower-category roads where AADT and fencing had no impact on UVCs. We conclude that wildlife fencing on roads characterized by higher speed and traffic intensity may decrease UVC numbers and at the same time shifting UVC occurrence towards roads characterized by lower speed and traffic intensity. Wildlife fencing re-allocates wildlife movement pathways toward roads with insufficient or no mitigation measures.

Railway mortality for several mammal species increases with train speed, proximity to water, and track curvature

Railways are a major source of direct mortality for many populations of large mammals, but they have been less studied or mitigated than roads. We evaluated temporal and spatial factors affecting mortality risk using 646 railway mortality incidents for 11 mammal species collected over 24 years throughout Banff and Yoho National Parks, Canada. We divided species into three guilds (bears, other carnivores, and ungulates), compared site attributes of topography, land cover, and train operation between mortality and paired random locations at four spatial scales, and described temporal patterns or mortality. Mortality risk increased across multiple guilds and spatial scales with maximum train speed and higher track curvature, both suggesting problems with train detection, and in areas with high proximity to and amount of water, both suggesting limitations to animal movement. Mortality risk was also correlated, but more varied among guilds and spatial scales, with shrub cover, topographic complexity, and proximity to sidings and roads. Seasonally, mortality rates were highest in winter for ungulates and other carnivores, and in late spring for bears, respectively. Our results suggest that effective mitigation could address train speed or detectability by wildlife, especially at sites with high track curvature that are near water or attractive habitat.

Are Roadkill Hotspots in the Cerrado Equal Among Groups of Vertebrates?

Understand the spatial distribution of wildlife roadkill is necessary to design mitigation measures minimizing damage to the fauna and the human population. Thus, we aimed to analyze the spatial distribution of wildlife roadkill in the Brazilian savanna ("Cerrado") to test whether roadkill hotspots match between the studied animal groups. We collected data of wildlife roadkill over a year in the southwest region of the state of Goiás, Brazil. To understand the distribution of roadkill on highways and to identify the aggregation hotspots, we used the modified two-dimensional Ripley K test and the two-dimensional hotspot identification analysis. We detected that birds and mammals have different aggregation points. These points may vary when the two groups are analyzed together or when species with greater abundance are removed from the analyses. Hence, we concluded that using generalist approaches including several species, are not enough, and can lead to erroneous conclusions. Therefore, it is necessary that the analyses be done in groups.

Identification methods of key contributing factors in crashes with high numbers of fatalities and injuries in China

OBJECTIVES

In China, serious road traffic crashes (SRTCs) are those in which there are 10-30 fatalities, 50-100 serious injuries, or a total cost of 50-100 million RMB (U.S.$8-16 M), and particularly serious road traffic crashes (PSRTCs) are those that are more severe or costly. Due to the large number of fatalities and injuries as well as the negative public reaction they elicit, SRTCs and PSRTCs have become of great concern to China during recent years. The aim of this study is to identify the main factors contributing to these road traffic crashes and to propose preventive measures to reduce their number.

METHODS

49 contributing factors of the SRTCs and PSRTCs that occurred from 2007 to 2013 were collected from the database "In-depth investigation and analysis system for major road traffic crashes" (IIASMRTC) and were analyzed through the integrated use of principal component analysis and hierarchical clustering to determine the primary and secondary groups of contributing factors.

RESULTS

Speeding and overloading of passengers were the primary contributing factors, featuring in up to 66.3 and 32.6% of accidents, respectively. Two secondary contributing factors were road related: lack of or nonstandard roadside safety infrastructure and slippery roads due to rain, snow, or ice.

CONCLUSIONS

The current approach to SRTCs and PSRTCs is focused on the attribution of responsibility and the enforcement of regulations considered relevant to particular SRTCs and PSRTCs. It would be more effective to investigate contributing factors and characteristics of SRTCs and PSRTCs as a whole to provide adequate information for safety interventions in regions where SRTCs and PSRTCs are more common. In addition to mandating a driver training program and publicization of the hazards associated with traffic violations, implementation of speed cameras, speed signs, markings, and vehicle-mounted Global Positioning Systems (GPS) are suggested to reduce speeding of passenger vehicles, while increasing regular checks by traffic police and passenger station staff and improving transportation management to increase income of contractors and drivers are feasible measures to prevent overloading of people. Other promising measures include regular inspection of roadside safety infrastructure and improving skid resistance on dangerous road sections in mountainous areas.