BRENNEN

Industries / Construction and property / Heavy civil

Paving

Paving builds and repairs the hard surfaces of roads, airfields, car parks and paths: a graded and compacted base of soil and crushed stone, topped with a layer of asphalt or concrete that carries traffic and sheds water.

c. 2500 BCE

Workers of Egypt's Old Kingdom laid slabs of sandstone and limestone, with some logs of petrified wood, along a road about 7.5 miles (12 km) long from the basalt quarry at Widan el-Faras to a quay on ancient Lake Moeris; the American Society of Civil Engineers dates it to 2500 to 2100 BCE and calls it the oldest surviving paved road in the world.

asce.org

29

Milestones achieved across 6 domains.

6

To watch for: announced, not yet confirmed outside.

12

Open deliverables. Nobody has shipped them.

In one minute

Where it started

The oldest surviving paved road, a track of stone slabs in Egypt's Faiyum, is about 4,500 years old. The modern trade took shape between 1867, when a steam roller was built to press road layers down, and the 1930s, when a machine first laid asphalt and R. R. Proctor's test set how densely soil should be packed. Concrete streets began in Ohio in 1891, and from 1949 slipform pavers laid concrete without side forms.

Where it stands

The Global Roads Inventory Project, which joins about 60 road datasets from 222 countries, counts 21.6 million km of roads, and where it knows the surface, 35% are paved, most of them in North America and Europe. In 2024 the United States had 2.80 million of its 4.01 million miles of public road paved, by the Federal Highway Administration's count, and on some of its highway jobs rollers map the stiffness of each layer, pavers record the temperature of the whole mat and radar maps the density. In eight African and Asian countries the World Bank surveyed, about 174 million rural people, out of a total population of over 470 million, had no good all-season road within 2 km.

What is open

Much of what decides how long a pavement lasts is measured on single jobs and never pooled: roller readings against core densities, mat temperatures against how the mix reached the paver, and how cold spots in a new mat wore years later. Other checks still rest on the eye or a rule of thumb, such as when a fresh seal can carry traffic and how a test box of concrete is ranked, and crack-sealing robots are scored only by the teams that build them. Public records, phone sensors, cameras and open AI models are enough to start on most of these jobs.

Domains

AchievedWatch forOpen

Advances to build with

4 research notes

Open in paving

3 of 12 open deliverables

Open · Grading and base

A national sulfate screen

A free map that flags where subgrade soils are likely to hold sulfate, built for the whole United States from soil survey and geologic map data, and scored against published sulfate test results from road projects.

Built with what exists

gSSURGO, the US Department of Agriculture's gridded soil survey, which records gypsum content by soil layer; the US Geological Survey's State Geologic Map Compilation; QGIS, or Python with GeoPandas; sulfate test results and project locations taken from state transportation research reports.

Credit · unclaimed

Open · Compaction

Roller pass counts from video

Software that counts how many times the rollers pass over each patch of a fresh asphalt layer, from video taken by a camera on a tripod outside the work area, draws a coverage map, and is checked against independent counts.

Built with what exists

a phone or action camera on a tripod; open detection and tracking software such as Ultralytics YOLO with the ByteTrack tracker, or Meta's SAM 2 for following objects through video; OpenCV to map the picture onto the ground from four measured points; Python.

Credit · unclaimed

Open · Asphalt paving

Transfer vehicles, retested at scale

A comparison, across at least 100 recent paving projects that have thermal profiler records, of how much of each mat was cold when the mix came from a material transfer vehicle, from a windrow elevator (a machine that picks up mix dumped in a row on the road) or straight from trucks.

Built with what exists

Profiler data exported from Veta or the vendors' software; the AASHTO PP 80 severity categories; project diaries or equipment lists obtained through public records requests; Python with pandas; a small extraction model such as Muse Ember (see the Note "The jobs agents repeat") to pull equipment lists from scanned diaries, checked by hand on a sample of 50.

Credit · unclaimed

Where it sits