The adoption of a safe system approach is seen by many as a critical element to achieving the aims of vision zero (i.e., eliminating fatalities and serious injuries on roads). Unlike in traditional approaches, in a safe system approach it is acknowledged that road users are humans who are prone to make errors and that the result of those errors should not be a serious injury or fatality. Accordingly, a safe system approach adopts a human-centric design philosophy whereby human fallibility and human vulnerability must be understood and integrated into the design of all elements of road infrastructure. Despite it being introduced more than two decades ago, efforts to integrate the safe system philosophy into national highway design guides have been limited. In fact, there is no information in design guides on the extent to which a road designed to meet minimum design requirements is able to handle driver demand. Furthermore, the safety impacts of meeting or deviating from recommended design standards are also unknown.

Despite this lack of information, roads are still built to meet standards recommended in design guides. Safety problems occur on those roads and are often addressed by introducing certain countermeasures, which sometimes include geometric changes to a road’s alignments. Although those countermeasures are often effective in improving safety on the existing highways, they do not address the root cause of the problem. In other words, the fact that deficiencies in recommended design standards might have contributed to certain safety problems on the existing highways is often neglected when new roads are designed. New roads are usually designed to meet the same standards as existing roads, which results in the same safety problems. This contradicts the core principles of a safe system approach where it is required that safety problems and the systems failure to accommodate driver demand are understood and integrated into the design process.

To address this problem, this thesis proposes the adoption of a performance-based design (PBD) approach whereby links between driver capabilities, safety performance, and geometric design elements on existing roads are first established when formulating design requirements for new facilities. One obstacle to the adoption of such an approach is the challenges associated with surveying information about geometric design elements of roads on a large scale. Therefore, the first phase of this thesis focuses on the development of novel algorithms that facilitate large-scale extraction and assessment of different geometric elements on highways scanned using mobile Light Detection and Ranging (LiDAR) technology. In particular, the first phase focuses on the development, testing, and validation of algorithms for (i) extraction and slope estimation of road cross-sections, (ii) the detection and the extraction of attributes of horizontal alignments, (iii) the inventory and clearance assessment of overhead assets, and (iv) the assessment of sight distances. The developed algorithms are fully automated and facilitate assessment of the aforementioned features along entire highway corridors in an efficient and accurate manner.

The second phase of the thesis focuses on conducting a performance-based assessment of stopping sight distance (SSD) requirements on highways. The performance-based assessment is conducted deterministically and probabilistically with the aims of (i) developing an understanding of the underlying links between demand for sight distance, geometric integrity, and safety performance on existing highways, and (ii) developing a framework for future studies interested in conducting a performance-based assessment of other geometric design elements. The assessment is conducted on over 40km of crash prone highways in Alberta where available sight distance is first quantified and then assessed against deterministically defined, and stochastically simulated driver demand. Among other findings, the assessment revealed that a significant proportion of the analyzed highways did not satisfy the SSD requirements of up to 70% of the driving population. These finding indicate the importance of adopting a probabilistic performance-based approach, which integrates driver capabilities and anticipated safety performance when designing new highway facilities.

In addition to facilitating performance-based assessment of highway geometric elements, algorithms developed in this thesis can be used for efficient network-level asset management as well as for the assessment of structural integrity of geometric elements on roads.