Industries / Construction and property / Engineering and architecture / Geotechnical engineering
Drilling and sampling
Drilling and sampling is the field work of boring into the ground, pushing probes through it and bringing up pieces of soil, so that engineers know what lies under a site and how strong it is.
5
Milestones achieved, from 1919 to 1992.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1919
A vane twisted in clay
John Olsson, secretary of the Swedish State Railways' Geotechnical Commission, designed what K. Rainer Massarsch and Bengt Fellenius call the very first vane borer "as far as can be determined", and first used it in 1919 while building the Lidingö Bridge near Stockholm. A vane borer is a rod with flat blades that is pushed into the ground and turned; Olsson's version measured how stiff the soil was around long piles. In 1948, at the second international soil mechanics conference in Rotterdam, Lyman Cadling (then Carlsson) presented the vane borer in its modern form, which finds the strength of soft clay from the twisting force needed to turn it.
Credit
John Olsson, Swedish State Railways; Lyman Cadling, Swedish Geotechnical Institute
c. 1923
A piston keeps samples whole
About 1923 John Olsson developed the first piston sampler, by Massarsch and Fellenius's account. A piston sampler pushes a thin tube into soft clay while a piston inside holds the soil in place, so the sample reaches the laboratory less disturbed than one that is hammered or scooped out. The Swedish Geotechnical Institute later built on it: Torsten Kallstenius received a doctorate in 1963 for an improved piston sampler, and the Swedish foil sampler, described in 1950, took very long unbroken samples.
Credit
John Olsson; Torsten Kallstenius and Walter Kjellman, Swedish Geotechnical Institute
1936
A cone pushed by hand
At the first international soil mechanics conference, at Harvard in 1936, P. Barentsen of Rijkswaterstaat, the Dutch public works agency, described a cone with a 60° point and a 10 cm² base on a rod inside a gas pipe. Workers pressed it into the ground in 50 cm steps and read the force needed from a pressure gauge, a quick and cheap way to map soft clay and peat lying on firm sand under new roads. This is the ancestor of the cone penetration test (CPT); Laurence Wesley and Michael Dobie date it to the early 1930s and write that Dutch engineers in Indonesia added a sleeve that measures side friction in the 1950s, and that electric probes with sensors in the tip have since replaced the mechanical cone in most places.
LessonWesley and Dobie warn that mechanical and electric cones, both still in use (the mechanical one widely in Indonesia), can give significantly different readings, so the type of probe must be known before the numbers are used.
Credit
P. Barentsen, Rijkswaterstaat
1983
Same test, different energy
The standard penetration test (SPT) drives a tube into the bottom of a borehole with a 63.5 kg (140 lb) hammer dropped 76 cm (30 in) and counts the blows needed for the last 30 cm; ASTM, the US standards body, calls it the most frequently used drilling test in subsurface exploration worldwide. In 1983 W. D. Kovacs, L. A. Salomone and F. Y. Yokel of the US National Bureau of Standards, in a report for the Nuclear Regulatory Commission, found that the energy reaching the drill rods ranged from 30 to 85 % of the standard, and that experienced operators lifting the hammer with a rope wound round a spinning drum (a cathead) did not keep to the 30 in drop from blow to blow. ASTM's current SPT standard, D1586, puts the range for today's hammers at 45 to 95 % and says blow counts are now commonly corrected to 60 % energy.
LessonThe same blow count from two rigs can describe different ground, so a hammer's energy has to be measured, not assumed.
Credit
W. D. Kovacs, L. A. Salomone and F. Y. Yokel, National Bureau of Standards
1992
One format for ground data
The Association of Geotechnical and Geoenvironmental Specialists (AGS), a UK trade body, published the first edition of Electronic Transfer of Geotechnical Data in March 1992. Its plain-text format, known as AGS format, lets a drilling crew, a laboratory and a designer pass borehole, test and monitoring data between different programs without typing them in again. A 2019 article in Geodrilling International calls it the standard not only in the UK but in America, Australia, Singapore and many other countries.
LessonThe same article warns that AGS checkers test only a file's structure, so a file can pass and still hold impossible values, such as a sample taken deeper than its hole or a core recovery of 1,000 %.
Credit
the Association of Geotechnical and Geoenvironmental Specialists
Watch for
A probe that pushes sideways
ROBOCONE is a device that fits on a standard cone penetrometer rod. At a chosen depth a module pushes out sideways to measure how the soil resists a sideways load, the kind of load an offshore wind turbine's foundation carries. The University of Southampton and the University of Bristol say they ran its first tests in clay in June 2026 at Onsøy, Norway, with the Norwegian Geotechnical Institute, and called the week "full of challenges and problem solving".
Achieved whenPublished data from the device agree with laboratory or full-scale load tests at the same site, and a company or agency other than its developers uses it on a project.
Claimed by
University of Southampton and University of Bristol, July 2026
Ground data across standards
The Open Geospatial Consortium (OGC), a standards body for map and location data, ran a Geotech Interoperability Experiment to write one common model for describing geotechnical data. The model is meant to bridge the existing formats for such data and link them to the buildingSMART standards used for digital building models. Its engineering report, by authors from France's geological survey BRGM, Germany's Fraunhofer IOSB and others, was posted in December 2025.
Achieved whenOGC adopts the model as a standard and a national data registry or an established format such as AGS publishes data that conforms to it.
Claimed by
Open Geospatial Consortium, December 2025
Open
01
Drop height from phone video
A phone app that films a standard penetration test and measures, from the video alone, the height of every hammer drop and the number of blows, then reports how far each test strayed from the standard 76 cm (30 in) drop.
The briefCredit · unclaimed02
Soil names from sample photos
A model that looks at photographs of soil samples from boreholes and proposes the soil's name and group symbol (for example "lean clay, CL"), as an engineer would write it on a borehole log, scored against both the engineer's field description and the laboratory's classification of the same sample.
The briefCredit · unclaimed
Open 01 · the brief
Drop height from phone video
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A phone app that films a standard penetration test and measures, from the video alone, the height of every hammer drop and the number of blows, then reports how far each test strayed from the standard 76 cm (30 in) drop.
Why it is still open
ASTM's D1586 still allows hammers lifted by rope and cathead. The National Bureau of Standards found in 1983 that experienced operators using them did not hold the standard drop; among its recommendations was a colored band painted on the guide pipe to help them aim. Measuring a hammer's energy needs a separate test method, ASTM D4633. Researchers have filmed the hammer: Hong Kong's Geotechnical Engineering Office, in a study that has gathered over 4,000 energy measurements since 2018, used a high-speed camera at up to 4,000 frames a second with graph paper stuck to the hammer and tracked its motion by hand in software. Automatic recording relies on sensors fitted to the hammer system, such as an apparatus Marl Technologies of Edmonton patented in 2021. This research found no tool that measures every drop automatically from ordinary phone video.
nvlpubs.nist.govstore.astm.orgissmge.orgpatentimages.storage.googleapis.com
Built with what exists
A phone filming at 240 frames per second; a printed scale target clamped beside the guide pipe; open-source tracking with OpenCV; a small image model to mark each blow.
First step
Build a bench rig that drops a 5 kg weight along a marked pipe, film 100 drops from known heights, and test the tracker on them.
Done looks like
On 20 tests run by a licensed drilling crew and filmed from a safe distance, video drop heights within 25 mm of frame-by-frame measurements made by two people for at least 95 % of blows, and blow counts that match the driller's log, with the videos and results published.
Open 02 · the brief
Soil names from sample photos
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A model that looks at photographs of soil samples from boreholes and proposes the soil's name and group symbol (for example "lean clay, CL"), as an engineer would write it on a borehole log, scored against both the engineer's field description and the laboratory's classification of the same sample.
Why it is still open
Deep learning already sorts photos of rock cores by rock type, as in a PLOS ONE study of ten rock types common in underground engineering, but soft soil samples are named by grain size and plasticity, not by rock type. The nearest published work on soil classifies loose excavated soil, not borehole samples. A 2023 system by L. T. Zhan and colleagues, in Computers and Geotechnics, which combines soil photos with a probe pushed into the soil, sorted excavated soils at transfer points into 12 types with 88.7 % accuracy, and a 2023 National University of Singapore prototype reached 85 % against grain-size tests on 493 samples. This research found no published comparison between a photo model and engineers' own field descriptions of borehole samples.
Built with what exists
A phone or fixed camera with a color reference card; a vision-language model or a fine-tuned image classifier; laboratory results (grain size, liquid and plastic limits) for the same samples.
First step
Ask a university soils laboratory for permission to photograph 50 samples already going through classification tests this month, and record their results.
Done looks like
On 500 photographed samples from at least three regions, each with a laboratory classification, a published table of how often the model's group symbol matches the laboratory's, set beside how often the field engineer's description does; photos and labels released openly.