Saturday, 31 August 2024

Construction Robots Onsite in 2024

 Some examples of robots now in use




Over the last decade there it has been suggested many times that robots will take over construction sites, but today they are conspicuous by their absence. So how much progress has been made in actually getting robots onsite in construction?


The International Organization for Standardization (ISO) defines a robot as ‘a programmed actuated mechanism with a degree of autonomy to perform locomotion, manipulation or positioning.’ There are separate definitions of industrial robots and service robots, with construction and agriculture robots classified as service robots because they are mobile. Not regarded as robots are software, remote-controlled drones, voice assistants, and autonomous vehicles. The International Federation of Robotics estimates there were 3.9 million installed industrial robots in 2023.


For the purposes of this post the ISO definition is followed, where a robot is an autonomous or semi-autonomous mobile machine that undertakes or assists with a specific task that can be found working onsite. Excluded are exoskeletons, software systems and platforms, offsite manufacturing, remote controlled equipment and drones, autonomous excavators and graders, and 3D concrete printing.


Included are 17 robots, three for bricklaying, five for layout printing and surveying, four for reality capture and site monitoring, and one for each of drywall finishing, anchor hole drilling, rebar fixing, underfloor insulation, and solar farm construction.


Also, by restricting this review to onsite robotic equipment currently in use startups are not included. There are many of these around the world and some have reached demonstration stage with their technology. However, getting a robot from the research and development stage to onsite use is a difficult path and few startups have so far been successful, and often has required partnering with a larger, more established firm.


There is one technology in particular where there is rapid development that will be important to future construction, and may be a transformational technology, that is not included here. These are the humanoid robots being developed by Boston Dynamics, Agility Robotics, Figure AI, UBTECH Robotics, Engineered Arts, Hanson Robotics, Sanctuary AI, PAL Robotics, SuperDroid and SoftBank Robotics. The Tesla Optimus robot is due in 2025 for Tesla factories and 2026 for sale. The Amazon Digit warehouse robot has been developed by Agility Robotics and is also due to be trialled in 2025. 


Bricklaying


Figure 1. Bricklaying robots


FBR

The Hadrian X is a truck mounted arm that builds walls while leaving window and door openings. It uses custom blocks about twice the size of a standard brick with glue applied to the bottom when placed. Developed in Australia, 10 projects have been completed in WA, and it has just been approved for site use in the US.


Monumental

Monumental’s robot bricklayer places bricks and mortar, operated by Atrium, an AI- software. Developed in the Netherlands, Monumental completed a 15-metre facade in 2023, has since done a social housing project, and now has partnerships with 25 general contractors.


Construction Robotics

From New York company Construction Robotics, the MULE (Material Unit Lift Enhancer) comes in two versions. It is lift-assist equipment designed to handle and place heavy material on construction sites, using a 3 metre arm attached to a stand. The MULE robot lifts and places bricks and blocks for a bricklayer as they lay courses. There are several case studies on their website, and it was used in Australia on the Sydney Metro project.


Drywall/Plasterboard Finishing


Canvas

The San Francisco company’s drywall finishing robots capture 99.9% of the dust produced by sanding, applies a layer of joint compound over wet tape, and comes in two models that telescope to 12 or 15.5 feet (3.7 and 4.6 meters). It can produce a paint ready surface in two days, one for applying mud and one for sanding, with Level 4 targeted spray and Level 5 finishes. Canvas uses the Formant platform for data collection, analysis, and robot management, and has over-the-air software updates. Canvas has a strategic partnership with Hilti.


Figure 2. Drywall and anchor hole drilling robots



Anchor Holes


Jaibot

Liechtenstein based company Hilti launched the Jaibot in 2020. A semi-autonomous, BIM-enabled, mobile robot it drills anchor holes for mechanical, electrical and plumbing services and has been used on projects around the world by major contractors like Skanska, Bouygues, De Groot etc.  Norwegian company nLink started developing the Drilly robot in 2013 before partnering with Hilti in 2017 to commercialise the technology as Jaibot.


Line Printers


SitePrint

The HP SitePrint robot is an autonomous machine that prints construction layouts from CAD files using different inks for different types of layouts. Wireless links the 2D CAD file to the robotic systems that set the control points on the site. SitePrint works with base stations from: Leica ICON Total Robotic Stations; Topcon Layout navigator; and Trimble RTS573. Used in Australia by Aptelia, who also have the TinySurveyorPlotter from Monsen Engineering, designed for road works and other hard surfaces.


Figure 3. Line printing and layout robots


 Dusty Robotics

Their FieldPrinter automates layouts and is integrated with Autodesk Revitt and autoCAD. Their customer stories page has 15 projects.  Example: On a hospital expansion project in Chicago by RG Construction, FieldPrinter was used to print the layout for the framing, mechanical, electrical, plumbing, and HVAC, with an estimated $300k in savings.


Civ Robotics

The San Francisco company backed by AlleyCorp has been used on over 100 US solar farms. It has two robots for coordinate collection and construction layouts. CivDot is an all-terrain 4 wheel drive rover that links to Trimble, Leica, Topcon and NTRIP base stations. It marks a point with paint but also has a speaker if a person is adding a physical marker like a flag or stake. The CivDash is for hard surfaces like roads and car parks. Example: Bechtel got a 6 times increase in coordinate collection for solar projects as CivDot gathered 125 points per hour while 2-person crews captured 20 per hour.


Tyker

German robot Tyker Plotter is a compact autonomous robot for marking lines on roads and asphalt using local sensors or a Total Station.


Surveying and Monitoring


Figure 4. Surveying and site monitoring robots


Nextera

Founded by MIT in the US, the Nextera DIDGE robot does site inspections and progress monitoring. Available as a tracked or wheeled robot it scans and documents a project. An AI powered system uses a variety of sensors to capture and compare wall, ceiling, mechanical, electrical and plumbing work to a BIM model.


DroneDeploy

The US company has AI driven ‘reality capture’ technology using a dock for aerial or ground robots that schedules inspections. The aerial DroneDeploy Flight App works on commercial drones and provides site mapping, facade and roof inspections. The aerial data can be combined with interior and exterior data from DroneDeploy Ground, which uses handheld 360 degree cameras, smart phones or Boston Dynamics Spotrobots that walk the site to collect data. Contractors like Turner Construction,Leighton Asia, and in Australia Hansen Yuncken have used these.


Onsite Technology is a Californian company that automates inspection and records the barcodes and location of panels in a solar farm using a rover with high resolution cameras. The geolocated data is uploaded to DroneDeploy to create a map of the panels and record their condition.


The Spot robot is also available with Trimble reality capture sensors. It can run autonomously once a site has been mapped. Used by contractors like Suffolk, Strabag, Virginia Tech etc.


Figure 5. SuperDroid robots


SuperDroid

US company SuperDroid has over a dozen robots designed for tasks such as site, pipe and crawl space inspection and materials handling. Capable of autonomous navigation after being driven once to create a map and path to follow. Used by Arup, Trimble, Balfour Beaty, DPR, Caterpillar, Gilbane, Black & Veatch etc. Superdroid have a quadruped robot called Cat that is like the Spot robot, and is developing a telepresence humanoid robot called Rocky for difficult or remote construction sites.


Rebar Fixing


Advanced Construction Robotics

The Pittsburgh company’s TyBOT is a rebar tying robot that self-navigates, self-ties, and does not require programming or a BIM plan. TyBOT’s partner, IronBOT, lifts, carries and places up to 5,000LB (2,268 kilos) an hour of rebar, from 9 to 60 foot bars (2.7 to 18.2 meters). ACR claims these robots can be set up and working onsite in a few hours. They have 2 dozen bridge and road projects on their website. Example: Over five months, D.T. Read completed the I-30 Arkansas Bridge using three TyBOT units that did 669,142 ties with a 30 percent schedule reduction.


Figure 6. Rebar fixing robots


Foam Insulation Spray


Q-bot

UK company Q-Bot robots can be controlled manually or operate autonomously to apply spray foam insulation to the underside of a suspended floor. Their 3D surveying platform has sensors to create point clouds that can be converted into simplified 3D models like Autodesk Revit and stored in AWS. Q-bot has seven UK and two French partners that use their robot.


Solar Installation


AES Maximo

AES Corporation is an American energy generation company, and they developed the Maximo robot for solar farm construction. It can install solar panels twice as fast as humans for half the cost. AES has installed 10 megawatts so far and plans to use Maximo to install five gigawatts over the next three years. Also used by Amazon constructing their largest solar farm at Bellefield.


Figure 7. Solar panel installation and underfloor insulation robots


Conclusion


Seventeen construction robots currently found onsite have been reviewed, however this is not intended to be comprehensive and no doubt there are others not included here. The ISO definition of a robot has been followed, which excludes exoskeletons, software, offsite manufacturing, remote controlled drones, autonomous excavators and graders, and onsite 3D concrete printing. Also excluded are startups with robotic equipment that is not yet being used onsite, and all the humanoid robots that are under development.


The degree of autonomy varies widely between these machines, but they have in common the ability to work without a human operator controlling their actions. Although remote controlled machines are becoming more common, like the Brokk or Husquavarna demolition machines, aerial drones like Skycatch or Voliro and tracked or wheeled rovers from Rugged Robotics or Acuity, the number of machines that can operate autonomously or semi-autonomously is still small. That said, many of these machines require humans to provide materials, like bricks for FBR’s Hadrian X, panels for Maximo and paint cans for line printers.


There are two broad categories of successful construction robots. The first is onsite data capture and use. This includes line printing, surveying, mapping, reality capture and inspection robots. These all have various combinations of sensors that allow autonomous site navigation, and range from relatively simple machines like line printers operating with a Total Station to complex combinations of technologies like Spot robots with site monitoring and mapping systems attached.


The second are repetitive tasks that have been automated successfully so robots can undertake them. These include bricklaying, drywall finishing, drilling, rebar tying, underfloor insulation and solar panel installing. The anchor hole drilling Jaibot has been available since 2020 and has been used on many sites around the world, making it arguably the most successful construction robot to date.


The important point is that these robots are now onsite. Although they may not yet be widely used, the fact that they are in use shows the construction industry is finding ways to develop and adopt them, despite the industry’s characteristics and structure, and the challenges and barriers faced by any major new technology. Further, the robots reviewed here are the first wave, leading the way for the wide range of robots and autonomous equipment that both established firms and startups are developing for the industry. In the near future more robots will be seen on construction sites, as many of the industry’s tasks become a working partnership between people and intelligent machines.





Friday, 16 August 2024

Investment in Physical and Intellectual Capital in 2023

 Australian capex in machinery and equipment, software and R&D 

 

 


The 2023 Australian System of National Accounts provided by the ABS includes data for industry investment in software, research and development (R&D), and machinery and equipment (M&E) [1]. Industry investment in physical and intellectual assets plays a vital role in economic growth through building capacity, upgrading technology, and increasing the productivity of workers.

 

This post compares the capital expenditure of 18 Australian industries in 2023 starting with M&E, then software followed by R&D, with industries ranked by expenditure. The ABS estimates of each industry’s net capital stock [2] are also included, with industries again ranked by expenditure. The third figure for each category compares each Industry’s share of total capital stock to the industry’s share of GDP. 

 

 

Machinery and Equipment 

 

With expenditure of $120 billion in 2023, M&E is by far the most important component of investment by Australian industry. In Figure 1 Australian industries are sorted by M&E capital expenditure in 2023, from lowest to highest. The two industries of Mining and Transport spent over $15 billion, Manufacturing spent over $10 billion, and Construction and Agriculture each spent around $9 billion. Those five industries accounted for 51 percent of total M&E capex, which however is more distributed than Software and R&D capex. The next five industries include three that spent between $6 and $7 billion, and two that spent between $5 and $6 billion. 


 

Figure 1. Industries ranked by machinery and equipment capital expenditure

 


Source: ABS 5204

 

Economic growth can come from either increased amounts of capital per worker or from technological progress and increased productivity. Since the financial crisis in 2006 the share of GDP of M&E capex has been falling, from eight to around four percent, and is now half the level it was before the financial crisis despite the decline in interest rates to 2021. With less investment the capital stock grows more slowly, leading to slower growth in output and therefore lower productivity. With a low rate of growth firms may not invest, or invest less, in expanding capacity and innovation (innovation was discussed in the previous post). The result is less economic dynamism and increasing economic inefficiency, leading to lower growth in productivity and GDP.

 

Very broadly, the net capital stock of M&E in each industry is around eight times their annual capex. Although a few industries like Electricity and Health move a couple of places, the ranking in Figure 2 generally follows that of annual expenditure. Agriculture and Construction swap places, but the top five industries are the same and they account for 45 percent of the total net capital stock. Those five asset heavy industries are by far the most capital intensive in Australia, particularly when compared with the bottom six service industries that only have between $10 and $25 billion in M&E capital stock. 

 

 

Figure 2. Net capital stock of machinery and equipment by industry

 


Source: ABS 5204

 

 

When the share of total capital stock for each industry is compared to its share pf GDP no real pattern emerges. Agriculture, Transport, postal and warehousing, and Rental, hiring and real estate services all have M&E capital stock shares that are much larger than GDP shares. Manufacturing and Electricity, gas, water and waste also have larger M&E shares. For Construction, Mining, Retail and Wholesale trades, and Accommodation and food services M&E shares are slightly higher than their GDP shares. On the other side are Health care and social assistance, Professional, scientific and technical services, Finance and insurance, and Education and training with M&E capital stock shares well below their GDP shares. 


 

Figure 3. Industry shares of M&E capital stock and GDP compared

 


Source: ABS 5204. GDP in current dollars at basic prices [3]. 

 

 

Intellectual Property: Software

 

Figure 4 shows industries ranked by capital expenditure on software, which is markedly different from the M&E rankings where Mining and Transport were the largest. Professional, scientific and technical services, which includes computer systems and IT services, with $5.7 billion had the biggest expenditure. Built environment related professional services like architecture, engineering, quantity surveyors and project management are also in this industry. [4].  


Finance and insurance with $3.7 billion and Information, media and telecommunications with $3 billion are the second and third largest. Three other industries spent over $2 billion, and those six industries accounted for 64 percent of the total. The top five industries accounted for 55 percent. Construction had the twelfth largest capex, and was sixth from the bottom in software capex. Total Software capex was $31 billion in 2023. 


 

Figure 4. Industries ranked by software capital expenditure 2023

 


Source: ABS 5204

 

The ranking of industries by software capital stock follows that for capex, with the exception of Manufacturing, which falls a couple of places to 12. Broadly, the value of software capital stock is a bit more than twice the value of 2023 capex, indicating a rapid depreciation rate for software of around three years. The top five industries in Figure 5 accounted for 57 percent of software capital stock. Construction remains at twelfth. 


 

Figure 5. Software net capital stock by industry 2023

 



Source: ABS 5204

 

The comparison of Software capital stock and GDP shares is dramatically different to the M&E shares. Here Professional, scientific and technical services, Finance and insurance, Information media and telecommunications, Transport, postal and warehousing, and Electricity, gas, water and waste have much larger capital stock shares than GDP shares. Agriculture, Construction, Mining, Manufacturing, Education and training, Accommodation and food services, and Health care and social assistance, have smaller shares. For Retail and Wholesale trades, Rental, hiring and real estate services, and Public administration and safety the shares are similar. 


 

Figure 6. Industry shares of software capital stock and GDP compared

 


Source: ABS 5204. GDP in current dollars at basic prices [3]. 

 

 

Intellectual Property: Research and Development

 

In Figure 7 industries are ranked by capital expenditure on R&D in 2023, from lowest to highest. Construction with $200 million is third from bottom and has the lowest R&D spend of any of the major goods producing industries like Agriculture, Mining and Manufacturing, which with $4.8 billion had the second largest investment in R&D. Professional, scientific and technical services, which includes the IT and computer services industries, had by far the largest R&D spend with $7.7 billion in 2023. There are five industries with capex above $1 billion, and those five accounted for 77 percent of all R&D expenditure.


 

Figure 7. Industries ranked by research and development capital expenditure

 


Source: ABS 5204

 

Capital expenditure on software is much more important than R&D for most Australian industries, the three exceptions where R&D was greater than software were Agriculture, Mining and Manufacturing. Some industries with low R&D capex are among the largest in software capex, such as Transport, postal and warehousing and Electricity, gas, water and waste. It is not uncommon for software capex to be many multiples of R&D, such four times more in Construction and five times more in Electricity.

 

The industry rankings change with the current value of R&D net capital stock, particularly for Agriculture and Mining where the value is relatively low. The leading seven industries are still leaders, Health is eight, Electricity is nineth and Construction is now at ten, ahead of Retail and Real estate services. The top five industries account for 76 percent of all R&D capital stock. 

 

 

Figure 8. Net capital stock of research and development by industry

 


Source: ABS 5204

 

The comparison of R&D capital stock and GDP shares highlights how concentrated R&D is in Australian industry. Only three industries have significantly larger capital stock shares: Professional, scientific and technical services, Finance and insurance, and Manufacturing. Three more have slightly larger shares, Information media and telecommunications, Education and training, and Public administration and safety. For all other industries their R&D capital stock share is less than their GDP share. In industries like Agriculture, Construction, Transport, and Retail the R&D shares are much smaller than GDP shares. 


 

Figure 9. Industry shares of R&D capital stock and GDP compared

 


Source: ABS 5204. GDP in current dollars at basic prices. 

 

 

Conclusion

 

Industry investment in physical and intellectual assets plays a vital role in building capacity and upgrading technology. Between the 18 industries the ABS provides data on the level of investment varied widely in 2023, with the share of capex of the top five industries increasing from 51 percent in M&E to 55 percent in Software to 77 percent in R&D. However, different industries are in the top five. 

 

In M&E capex in the top five industries Mining and Transport, postal and warehousing have much larger capex and capital stock than the next three industries of Manufacturing, Construction, and Agriculture. Policies to increase M&E investment could target those industries, although because M&E capex is more distributed than Software and R&D capex across the leading dozen industries a more general approach has traditionally been taken.

 

 In Software the largest expenditure was by Professional, scientific and technical services, Finance and insurance, Information, media and telecommunications, and Public administration and safety. R&D capex is highly concentrated in a few industries. Professional, scientific and technical services, Manufacturing, and Finance and insurance had the largest expenditure. For Software and R&D, capex policies that target the top three or four industries would be most effective. 

 

Economic growth can come from increased capital per worker or from technological progress and increased productivity. With investment the capital stock grows, and a low level of investment means slower growth in output, lower productivity, less economic dynamism and increasing economic inefficiency. 

 

The net capital stock of M&E in each industry is around eight times their annual capex, and ranking generally follows that of annual expenditure. The ranking of industries by software capital stock also follows that for capex, with the exception of Manufacturing, and the value of software capital stock is a bit more than twice the value of 2023 capex. industries with low R&D capex are among the largest in software capex, such as Transport, postal and warehousing and Electricity, gas, water and waste. It is not uncommon for software capex to be many multiples of R&D, such four times more in Construction and five times more in Electricity. Unlike M&E and software, the R&D capex industry rankings change for net capital stock, particularly for the low capital stock industries of Agriculture and Mining. 

 

For M&E net capital stock the top five asset heavy industries account for 45 percent of the total, and are very capital intensive compared with many service industries. The top five industries in software capital stock accounted for 57 percent.  The top five industries account for 76 percent of all R&D capital stock. The comparison of R&D capital stock shares and GDP shares highlights how concentrated R&D is in Australia, with only three industries having significantly larger capital stock shares.

 

The industry Professional, scientific and technical services includes computer systems design and has the highest expenditure on both software and R&D, has the largest capital stock, and a much larger share of total capital stock than GDP for them. For Australia this is the leading industry for intellectual property software and R&D investment and capital stock. Because this is the industry that includes scientific research and IT systems and services this is not surprising, but the wide gap between this industry and all the others suggests it has a specific and special role in the economy, and industry policy should reflect that.

 

 

 

 

[1] The data used here is from Tables 63 and 64 of ABS 5204, which comes out twice a year in February and October. All figures in this post are in current dollars, but the publication includes constant dollar estimates for expenditure since 1960. 

 

[2] Gross capital stock values each asset in use at the current price, Net capital stock is the written down value of gross capital stock. The difference between the net and gross value is accumulated depreciation.

 

[3] The basic price is the amount retained by the producer in respect of the good or service that is produced as output, minus any tax payable (including deductible value added taxes) plus any subsidy receivable. 

 

[4] Professional, scientific and technical services include scientific research, architecture, engineering, computer systems design and related services, law, accountancy, advertising, market research, management and other consultancy, veterinary science and professional photography.