Technology to enhance signalised intersection traffic operations

iMOVE’s Improved Sensing for Signalised Intersections project, conducted by the Queensland Department of Transport and Main Roads (TMR) and Queensland University of Technology (QUT) has been completed.

Objectives

The objective of this project is to explore various traffic detection technologies applicable to signalised intersections. The primary focus is on understanding the detection, sensing, and perception requirements for strategic management and tactical control of signalised intersections.

Through desktop reviews, surveys and interviews, this study documents the current national and international offerings, highlighting the strengths and limitations of available technologies for the selected applications.

Methodology

Research for the report was conducted by review of documents available online and shared by practitioners and vendors, and insights from practitioners gained via surveys and interviews. A systematic five step approach was employed.

Insight workshop with TMR: Gain insight into the work to date via a knowledge sharing workshop with TMR. This step scoped the aspects to be considered for Traffic Management Systems (TMS) applications, road users, and relevant sensing and perception parameters.

Literature review: Undertake a comprehensive review of scientific and public project reports from various scientific journals, databases, and websites to gather existing knowledge and findings.

Technology exploration and evaluation. Available resources were explored and evaluated to identify and review key technologies of interest with focus on detection, sensing and perception capabilities and complexities.

Practitioner and vendor engagement. Practitioners, prominent vendors, consultants, and academics were identified, surveyed, and/or interviewed to obtain detailed product information and insight from their field trials and case studies.

Synthesis of knowledge. This step involved synthesising insights gained from use cases, implemented case studies, and feedback from practitioners and vendors to understand the operational mechanics and suitability of technologies for various signalised intersection applications.

Report findings

The report yielded several key findings. Some of the key points were:

Non-intrusive traffic detection technologies are affected by weather and surrounding conditions. The extent of this impact varies depending on the site and technology. However, their inconsistent performance compared to traditional loop detectors, has limited their widespread adoption.

No single technology is universally suitable for all applications, and while deploying multiple technologies can enhance detection capabilities, managing such systems can be challenging. A promising solution is the use of hybrid sensors that integrate multiple technologies.

The choice of the deploying the appropriate technology depends on several factors, including site conditions, desired sensor outputs, usability, budget, and maintenance considerations.

Practitioner opinions

Despite the array of new technology available, loop detectors were found to be the preferred choice for practitioners. Reliability, established operational processes, practitioner experience and cost were cited as contributing factors for this preference.

Practitioners highlighted a gap between the data insights provided by new and emerging technologies and their current tactical and strategic use. For instance, while video and LiDAR sensors can provide real-time data on vehicle classification, they are not widely utilised, suggesting that the full potential of these technologies is not being realised.

Transitioning away from loop detectors will take time, but new technologies are developing rapidly and offer several benefits. Improvements in cost, accuracy, reliability, and efficiency are expected as these technologies evolve.

Technology – Potential and limitations          

Each technology offers specific advantages for traffic management, but the report also identifies limitations and areas where additional research is needed.

Amongst findings, video sensors beneficially offer both visual feed and traffic detection capabilities, but are prone to inaccuracies arising from weather conditions, illumination levels and projection issues as well as sensitivity to AI algorithm and calibration.

Radar sensors showed impressive accuracy in replicating loop detectors. However, because they use tracking algorithms to detect stopped vehicles, their accuracy decreases in stop-and-go traffic and pedestrian waiting areas.

LiDAR sensors offer the most comprehensive dataset but have a relatively short, warranted lifespan and higher deployment costs compared to other technologies

Hybrid sensors provide integration of multiple technologies into a single unit, offering a broad range of supported operations. However, they are not widely tested and the lifespan of these hybrid systems is not yet fully understood.

Practitioners found the flexibility of non-intrusive technologies beneficial for detecting active transport mode users. The research showed that almost all non-intrusive technologies were more effective at detecting cyclists, with video and LiDAR technology also proving promising for pedestrians in waiting areas. However, occlusion challenges are a concern for all non-intrusive sensors, making the number and positioning of sensors a primary consideration for practitioners.

Cost and maintenance

The selection of technology is influenced by both the initial product cost and the maintenance cost. The non-intrusive nature of the new technologies provides benefits in terms of maintenance requirement compared to loops. For example, most practitioners reported that radar sensors have lower maintenance requirements compared to other technologies. Similarly, video and LiDAR systems primarily require periodic lens cleaning every 6-12 months, often with minimal or no road closures.

While evaluating the whole-of-life costs for these technologies was beyond the scope of this research, interviews and surveys revealed that decision-makers perceive loop and video technologies to be the most cost-effective. LiDAR is seen as the most expensive option, partly due to its high product cost and limited deployment. Additionally, non-intrusive sensors are susceptible to vandalism, complicating overall cost estimates.

There is also a trade-off between cost and application accuracy; for instance, although LiDAR is costly, it provides the most detailed data which can be used for enhancing safety and performance. Similarly, deploying multiple video cameras focused on specific lanes can enhance accuracy.

Recommendations

Based on the findings of this report and advancements in technology, it is recommended to undertake comprehensive field trials of technologies (including hybrid sensors) for qualitative evaluation under different geometric and environmental conditions.

From this work, the authors also flag the need for detection performance standards that are tailored to different signal usage. Current standards (such as MRTS204) are primarily focused on motorway use and lack comprehensive guidelines for road user detection at signalised intersections.

Expected project impacts

The findings from this research have significant implications for both practice and academia:

1. Impact on Practice:
   • Strategic Focus for TMR: The insights direct practitioners, particularly at TMR, to strategically focus on hybrid sensors. This approach is vital for adopting alternative technologies that will shape the future of traffic signal control.
   • Empowering Flexible Solutions: Leveraging these findings allows TMR to explore and implement more flexible, adaptive solutions for traffic management, enhancing efficiency and responsiveness.
2. Impact on Academia:
   • Enhancement of Learning and Teaching at QUT: The findings are contributing to the enhancement of Learning and Teaching at QUT, where applied research in traffic operations is being integrated into the curriculum, benefiting students and educators alike.
   • Driving Innovation: The research is not only informing the next generation of transport engineers but also driving innovations in the field of traffic management, fostering a culture of continuous improvement and advancement.