Industries / Construction and property / Landscaping and lawn care
Planting and building
Putting in the plants and lawns, and building the hard parts of a garden: paths, patios, retaining walls and edging.
4
Milestones achieved, from 1953 to 2023.
1
To watch for: announced, not yet confirmed outside.
3
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1953
Seed sprayed in slurry
Hydroseeding mixes seed, fertilizer and water in a tank and sprays it onto bare ground. By Finn Corporation's account, it began in 1953 as a way to seed highway roadsides; a 500-gallon HydroSeeder in 1958 made it practical for landscaping companies, and from 1961 a wood-fiber mulch from International Paper was added to the mix to hold water and protect the seed. By 1970 one-pass seeding was established practice.
Credit
Finn Corporation; International Paper (mulch)
2006
Deep planting measured
The root flare is where a tree's trunk widens into its roots, and it should sit at the soil surface. Wells and colleagues planted red maples and Yoshino cherries with the flare at grade, 15 cm (6 in) deep and 31 cm (12 in) deep. Two years on, half of the deep-planted cherries had died and all at grade had lived; after three years, girdling roots circled 14% of the trunk on maples at grade, 48% at 15 cm and 71% at 31 cm. In 2014 the American Standard for Nursery Stock (ANSI Z60.1), which sets how nursery trees are sized and sold, clarified how the root flare should relate to the soil line.
LessonA tree set too deep looks fine on the day it is planted; Ohio's Department of Natural Resources reports Minnesota research finding that 73.3% of lindens lost in 1998 windstorms broke where buried roots had squeezed the trunk.
Credit
Wells, Townsend, Caldwell, Ham, Smiley and Sherwood (Arboriculture & Urban Forestry, 2006); AmericanHort (standard)
2018
Failed walls tallied
Robert and George Koerner of the Geosynthetic Institute collected 320 failed retaining walls built of compacted soil held by layers of plastic mesh (called reinforced soil walls). 75% were faced with concrete blocks. 76% had poor to moderate compaction, 73% were backfilled with silt or clay, and water inside or next to the wall played a part in 63%. 77% failed within four years of being built.
LessonWalls fail because of the soil and water behind them, not the blocks on their face, and most failures show up within four years.
Credit
Robert M. Koerner and George R. Koerner (Geotextiles and Geomembranes, 2018)
2023
Robot builds a stone wall
A dry stone wall is stacked without mortar, so each stone must be chosen to fit. A team at ETH Zürich led by Ryan Johns fitted a 12-ton excavator with laser scanners and planning software that scanned each boulder or chunk of waste concrete and worked out where it would fit. It built a 10 m freestanding wall from 109 pieces and a 65 m retaining wall from 938, including at a public park. AAAS reports the machine took on average about 1.2 minutes longer per stone than a group of people.
Credit
Ryan Johns and colleagues, ETH Zürich (Science Robotics, 2023)
Watch for
Paver machine grips by itself
Optimas, a German maker of paver-laying machines, says the 2026 upgrade of its PaveJet S24 picks up and sets whole layers of pavers from the pallet with an automatic grip-and-place function, cutting work for the driver, and that with good site organization it can lay in 20-second cycles and more than 1,000 m² (10,000 ft²) a day. Both figures are the maker's own.
Achieved whena contractor or trade magazine times the machine on a real job and publishes square meters laid per day and the share of joints that needed hand correction.
Claimed by
Optimas Maschinenfabrik H. Kleinemas GmbH, 2026
Open
01
Paver joint gauge from photos
A phone method that measures, across a whole patio, the gap between pavers (joint width) and the step in height between neighbors (lippage), from a set of photos, and marks the spots outside the job's tolerance.
The briefCredit · unclaimed02
Robot sets a paver pattern
A small robot arm that takes pavers from a stack and sets them one by one in a herringbone pattern on a leveled sand bed, keeping joints even, as a person does on a garden path.
The briefCredit · unclaimed03
Root flare check by photo
An image model that looks at a phone photo of the base of a young tree and says whether its root flare is buried, and by about how much, so a landscape crew or city can find deep-planted trees before girdling roots form.
The briefCredit · unclaimed
Open 01 · the brief
Paver joint gauge from photos
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A phone method that measures, across a whole patio, the gap between pavers (joint width) and the step in height between neighbors (lippage), from a set of photos, and marks the spots outside the job's tolerance.
Why it is still open
Checking is done with a ruler, a straightedge and the eye. The closest published work measured height steps between 5 m concrete slabs on an airport apron with a tripod laser scanner (Barbarella and colleagues, 2017), with fitting errors of about 1 to 3.5 mm. A web search for paver joint or lippage measurement by phone, photogrammetry or scanner found only road and airport work, and an arXiv search for pavers and robots or scanning found nothing.
Built with what exists
A recent phone camera; COLMAP or OpenMVG photogrammetry (building a 3D model from overlapping photos); Open3D to fit planes to each paver; a printed scale marker; a digital caliper and feeler gauges for checking.
First step
Lay a 1 m by 1 m patch of pavers on sand, measure 50 joints with a caliper, and photograph it from 30 angles to build the first 3D model.
Done looks like
10 patios or test patches, 200 joints and 200 paver edges measured by hand and by the method, with the mean and worst error in millimeters published for joint width and lippage; the method is useful if it stays within 1 mm.
Open 02 · the brief
Robot sets a paver pattern
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A small robot arm that takes pavers from a stack and sets them one by one in a herringbone pattern on a leveled sand bed, keeping joints even, as a person does on a garden path.
Why it is still open
Paver machines such as Optimas's lay whole factory-arranged layers with a driver, and patterns, curves and edges are still set by hand. ETH Zürich's excavator set boulders, not pavers. An arXiv search of abstracts for pavers or paving stones and robots returned no papers; Crossref, web and GitHub searches found brick-laying robots and a 2021 student control page for a "stone paving robot" with no description of the robot, but nothing tested at garden scale.
Built with what exists
An SO-101 open-design arm with LeRobot (see the Note "No download"); a control model trained from demonstrations with NVIDIA GR00T 2 (see the Note "Five demonstrations per task") or Runway Praxis-1 (see the Note "Open robot model weights"); a wrist camera; scale-model pavers, 3D-printed or cut from wood, light enough for a hobby arm; a 40 cm sand tray.
First step
Print 120 scale pavers, build the sand tray, and record 50 remote-controlled placements with LeRobot.
Done looks like
20 filmed runs of 50 scale pavers each, with a published tally of pavers set within 1 mm of the target joint width and level with their neighbors, pavers dropped or knocked, and seconds per paver against a person.
Open 03 · the brief
Root flare check by photo
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An image model that looks at a phone photo of the base of a young tree and says whether its root flare is buried, and by about how much, so a landscape crew or city can find deep-planted trees before girdling roots form.
Why it is still open
Deep planting is well documented as a cause of tree loss, and the check is still a person scraping soil away from each trunk. An arXiv search of abstracts for "root flare" returned no papers, Crossref searches found field studies of planting depth but no image method, and a GitHub search found only a 2013 tool that predicts root flare size from crown width.
Built with what exists
A phone camera; Label Studio for labeling; an Ultralytics YOLO classification model fine-tuned in PyTorch; a hand trowel and ruler to measure the true soil depth over the first large root.
First step
With a city forester's or property owner's permission, photograph 30 young trees from four sides, then measure the soil depth over each root flare.
Done looks like
300 labeled trees planted in the last five years, each photographed and then measured by hand, and a test on 100 more trees from a different town, with a published count of buried flares (more than 5 cm of soil) found and missed.