|Abstract:||This paper explores the current state of connected and autonomous vehicles (CAV) in the United Kingdom (UK), with a focus on positioning capability in rural areas. We carried out an initial investigation into UK CAV research and development projects in 2017 which highlighted a strong focus on urban CAV testing; however, it is rural populations who have the most to gain from CAV implementation. Rural areas experience the greatest numbers of road fatalities, poor public transport services, and low levels of digital connectivity. This study’s preliminary findings show that there is a distinct need to investigate the challenges that CAVs will face in the rural environment. This paper assesses whether there is a need to test CAVs rurally through the identification of these challenges and suggests avenues for solving them to assist with UK-wide CAV implementation. Addressing these challenges would also be paramount for the development of the UK’s automated agriculture industry, as farmland makes up much of the rural UK. To identify how the challenges uncovered in this investigation relate to different environments, definitions for the terms ‘rural’ and ‘urban’ are required. Although definitions exist, such as the Rural Urban Classification (RUC) in the 2011 Census, which sub-divide what is urban and rural into more distinct, descriptive clusters, these are purely population-based and cannot be referenced viably when describing the environment of a CAV or any other road vehicle. This paper determines new, more appropriate, road-based definitions for CAV environments to which the challenges uncovered can, and do, appropriately relate to. These definitions are adapted from the RUC sub-divisions but are primarily based on road type and quality. Based on photographic evidence collected during this experiment the new definitions developed for urban/rural roads are applicable, although this validation is based on subjective observations. Across the UK, the quality and consistency of road infrastructure and signage varies, as does the availability of network connectivity. To ensure the most extensive positioning capabilities are achieved it is suggested that CAVs use a combination of local on-board sensor technology and GNSS satellite positioning technology. In the UK, networks of static GNSS reference receivers enable CAVs to utilise real-time kinematic (RTK) GNSS positioning across the country. Depending on availability, network RTK can deliver the high positioning accuracies required for the safe navigation of CAVs. This paper’s primary investigation determines how NRTK availability and reliability vary across urban and rural environments by analysing positioning data from three road test experiments in and around the city of Nottingham, UK. These tests traverse a variety of typical UK environments and road networks to highlight the varying nature of NRTK positioning across these environments. The qualitative categorisation of these environments in terms of satellite positioning capability, in relation to the overhead cover density, enables the numerical analysis of positioning success to determine which environments lack effective positioning capability. The investigation goes on to address the hypothetical capability of on-board positioning equipment; determines how infrastructure, signage and road surface conditions vary across environments; and then explores the associated effects on localised positioning. The relationships between satellite and on-board positioning techniques are then assessed and their associated and combined challenges identified. Whilst contradictory results exist between the three tests the overarching result highlights that, whilst CAV positioning capabilities face similar challenges on both urban and rural roads, CAVs on rural roads are more likely to experience blind-time, whereby the vehicle cannot autonomously position itself to the required levels of safety. This is in-part due to the loss of NRTK positioning (95% of time spent under dense rural tree cover resulted in lost positioning) combined with the lack of effective road markings. These findings suggest that UK roads are not yet ready for the implementation of CAVs and highlight challenges specific to rural roads; an area of study that is currently neglected. We conclude that there are two key challenges that need to be addressed. The first challenge is the issue of GNSS positioning accuracy, which this investigation finds to be primarily governed by the line-of-sight connection between satellite and CAV GNSS receiver. Whilst greater proportions of urban environments tend to be more densely covered, the average positioning fix loss is 26% and 25% of the time across urban and rural environments respectively. Whilst rural tree cover is less common than urban building cover, the nature of overarching trees above the roads suggests that they play a greater role in disrupting GNSS connections. The availability of NRTK needs to be developed to reduce the loss of position fix in these densely covered environments and more work on the specific effects of tree cover is needed. The second challenge is the lack of physical and digital supporting infrastructure that CAVs use. This infrastructure is required to improve vehicle guidance and enhance road safety. With 18% of rural roads being ‘unreadable’ by CAV on-board sensor technology, and with a well-documented lack of UK cellular coverage, this paper does not deem it safe at present for CAVs to be implemented on UK rural roads. Cellular coverage and the quality of physical road infrastructure and markings not only need to be improved, but also frequently maintained, to ensure CAVs have a consistent and safe environment in which to operate.|
Proceedings of the 2019 International Technical Meeting of The Institute of Navigation
January 28 - 31, 2019
Hyatt Regency Reston
|Pages:||828 - 842|
|Cite this article:||
Walters, Joseph G., Meng, Xiaolin, Xu, Chang, Jing, Hao (Julia), Marsh, Stuart, "Rural Positioning Challenges for Connected and Autonomous Vehicles," Proceedings of the 2019 International Technical Meeting of The Institute of Navigation, Reston, Virginia, January 2019, pp. 828-842.
ION Members/Non-Members: 1 Download Credit