Motivation

Commercial Context and Background – rural New Zealand

Why do road defects matter?

Most countries around the world use a reactive road maintenance regime i.e. they fix road defects once they start impacting on safe travel. OECD countries (OECD, 2019) spend on average 1-2% of GDP on maintaining roads and related infrastructure. A preventative road maintenance regime shows the potential to meet growing loads on roads (by kilometres travelled) with shrinking budgets for local transport authorities (MOT, 2019). This paper proposes an affordable method to assist to shift from reactive to preventative road maintenance with similar or even smaller budgets.

How do road defects occur?

The genesis of road defects can be explained as a simple erosion problem. Typically, a road defect on a paved road happens when pressure/ load is applied to the road. An average truck & trailer unit has the impact of up to 30 cars travelling the same road. If you have changing weather conditions with substantial precipitation, frost, the growing pressure from above and moisture getting into the substrate under the road surface start the erosion process. Dependent on the soil type and substrate below the road, erosion sets in. When cracks form on the surface, the pressure from above starts being accentuated and road defects grow. There are a multitude of manifestations of road defects.

NZ-un-sealed-road-defects-01

Figure: sample road defects (sealed/ unsealed) in New Zealand

Most people are familiar with potholes. These are openings in the road surface of around 30 centimetres in diameter and several centimetres deep. However, there are other road defects which can be even more dangerous to safe driving like edge break. This is a specific problem on tight rural roads with heavy commercial vehicle traffic. Transversal and longitudinal cracks, heaving (lifting of the surface), scabbing (reduction of friction of surface material) and early warning signals like alligator cracks are only some of the manifestations of road defects on paved roads. On unsealed roads in rural areas rutting, wheel tracks, wash outs are other relevant road defects. The impact of unsmooth roads can be substantial.

What are the impacts of road defects?

Road defects can cause serious crashes by drivers swirling around a pothole, heaving, edge break. They can brake axles, reduce travel comfort, slow down traffic, increase friction and therefore fuel consumption and typically lead to higher repairs and maintenance. In Northland, the biggest region in New Zealand, a truck operator can spend up to $7,000 per year per truck more on repairs and maintenance due to the bad quality of the roads.

If a trailer gets close to the edge on these narrow roads and the edge has eroded away, the trailer starts moving backwards and forwards. In most cases the momentum is too much for the driver to correct and result in the truck and trailer unit ‘rolling’ i.e. tipping over. In New Zealand a minor crash has $104,000 social cost associated with it, a serious crash $923,000 and a fatal crash $4.9 million (NZTA, 2018). This is not including the business cost for the vehicle operator. In the past the described rural road, Otaika Valley Road (OVR), has been closed multiple times for a whole day when a truck related crash occurred. Other vehicles needed to get re-routed. In several instances additional crashes occurred on the recommended detour.

The minimum business cost to a truck operator is $80,000 even with only minor material damage but can amount to $600,000. For small operators this can be the difference between being in business or having to declare bankruptcy. The realities of road defects for those affected parties are enormous. That is another important reason why preventative road maintenance is critical on those high frequency rural roads to keep those roads safe. Affordable and near real time road defect detection is an important step towards this goal.

How do we detect road defects at present?

In most countries around the world trained road safety engineers visually patrol the road network regularly. This means driving at low speeds (40km/h or less), stop when they see a road defect, manage the traffic flow and mark & record the road defect. This can be a time consuming and a not very precise, subjective undertaking. Other methods are automated road scans using sophisticated equipment like multiple laser scanners, high pressure water reflection methods like in the SCRIM method. Recently vehicle based inertial momentum (IMU) and video analytics have been introduced. Some with very high and some smartphone-based technologies with lower accuracy.

In our paper with KIT we proposes a very efficient hybrid method of IMU and video analytics.

An example – Otaika Valley Road (OVR, Whangarei, New Zealand)

We like to put the above into context of the specific case of a rural road in Northland in New Zealand. Otaika Valley Road (short: OVR) is a 12 km long rural road which strategically connects several forests to the Port of Whangarei (Northport). Every day about 10,000 tonnes of radiata pine wood are felled and transported from the forests to the port for export. Usually a truck and trailer unit travels empty to the forest and comes back fully loaded to about 50 tonnes.

OVR-CAS-simple-complex-04

Figure: Otaika Valley Road (OVR) South of Whangarei with NZTA Crash Information

In the 1960’s OVR road was a horse track and was never designed to take the current 700+ vehicle movements per day. The road is very narrow. Over the last ten years over 90 crashes occurred on this road, with three fatal accidents.

OVR-CAS-02

Fig 2: Crash Statistic on Otaika Valley Road (NZTA, 2019)

It takes two human road inspectors about 3. 5 days to map all road defects on this road. Then another half day to review their findings and transfer the data into the nationwide road asset management system (RAMM). In Northland alone, with over 5,000 km of roads (including 3,000 km unsealed roads), to monitor the road network can be very time consuming if it must be done to a good quality standard.

This is where the proposed semi-automated method of IMU and video analysis comes into play. A vehicle with the technology as payload can travel the road network as part of their daily work routine and can map 600 km or more every day per vehicle. In Northland N3T uses logging, milk and delivery trucks to take the monitoring equipment. The data from these vehicles gets correlated and analysed in time and spatial dimensions. This gives a very accurate status of the road defects on the network in near real-time.

The four-day data collection, analysis process by human road inspectors can be semi-automated and done in less than 20 min for a 12 km rural road like OVR.

To understand OVR better, here is a link to a VR 3D flyover  (open with Microsoft Films & TV or similar 3D capable app) of the road with the last ten years road defects marked up. This has been built with the Unreal Game Engine and is based on Northland Regional Council Lidar Survey data.