Construction is an important and labor-intensive industry with a complex working environment. In order to handle the challenges of COVID-19, an aging population, and the accelerating pace of life, it is urgent to develop advanced technology used in the construction industry [1]. In construction, next to water, reinforced concrete is the most commonly used material worldwide and has a profound impact on construction quality [2,3]. As an effective tool to increase efficiency in the building process and reduce casualties among construction workers, various concrete construction robot has been proposed and applied in the field, which provides encouraging capabilities.

The early research and practice of concrete construction robots in the construction industry can be traced back to the 1970s and 1980s, and the development of new technologies to improve productivity in the construction industry has been envisaged since the late 1980s [4]. With the development of artificial intelligence technology, sensor technology, and BIM technology [5], the construction robot technology for site construction has gradually matured [6,7], and a large number of single-task construction robots and integrated robots are emerging at construction sites, making construction tasks more efficient, accurate and safe [8,9]. With the maturity of autonomous driving technology and laser detection technology [10], more concrete construction robots have applied the technology of autonomous driving detection and sensor detection. These make the building robots more intelligent and construction more precise.

Although considerable research has been carried out and various types of structures and technologies have been developed [11,12], the field is far from mature, and the use of concrete construction robots is still limited [13]. With the popularization and application of concrete construction robots [14,15], a series of problems, such as the adaptation of robots to complex building environments [16], the management and maintenance of robots at concrete construction sites, and the training of new industrial workers, are worth in-depth discussion. Meanwhile, technical research on concrete construction robots is scattered and lacks unified technical systems and standards. Therefore, this paper sorts out and introduces the research cases of concrete construction robots at home and abroad and puts forward corresponding development suggestions according to the use of concrete construction robots. This could broaden the research ideas for related enterprises and institutions and provide a reference for the development of concrete construction robots.

According to the application situation, existing concrete construction robots can be categorized as concrete distribution robots [17,18,19], concrete leveling and compaction robots [20,21,22], concrete floor finishing robots [23,24,25], concrete surface painting robots [26,27], 3D printing construction robots for additive manufacturing [28,29,30], and construction-used surveillance robots [31,32,33]. The common concrete construction process is shown in Figure 1. A large variety of autonomous vehicle driving and monitoring techniques [34,35,36] has been widely used in concrete construction robots; however, they lack clear relevance to the key functions in the building process. Based on this point, we attempt to explain the basic functions of concrete construction with the related autonomous robots. Due to the complexity and diversity of the building concrete construction environment, the current workload that can be handled by concrete construction robots is far from meeting actual engineering needs, indicating that the global concrete construction robot application market is still in the incubation period.

Recently, advancements in the application of 3D printing in building construction have developed rapidly in the past decade [101]. Contrary to the traditional cutting material construction method, only the necessary materials are provided, which greatly reduces waste. Using the concept of 3D printing to superimpose and pour concrete layer by layer after slicing, the building structure can achieve the primary molding of the building body [102]. However, the current 3D printing process is expensive and is not suitable for large-scale products and traditional design methods. The range of materials that can be used is very limited. Thus, the construction cost of 3D printing is high.

A typical 3D printer and its delivery system are shown in Figure 11. In order to ensure the implementation of 3D printing technology, the first step is to design a construction path through 3D to 2D slicing software. It starts by cutting the 3D shape of an object into thin, flat layers of a certain thickness and then designs the path of each layer. All of these paths consist of contours and fill patterns [103]. At present, the three 3D printing methods that are used in construction generally include contour crafting (CC), concrete printing, and D-shape 3D printing [104]. Meanwhile, a number of new printing methods are being investigated, such as rock printing [105], inkjet powder printing [106,107], multi-story apartment printing, and complete office printing.

The basic principle of CC technology is the same as that of other ordinary 3D printers. The material is stacked layer by layer in the specified position using extrusion equipment [108]. The difference is that the printing material of CC is high-density and high-performance concrete. The CC machine (Figure 12) developed by the University of Southern California consists of an XYZ gantry system, a nozzle assembly with three motion control components (Figure 13) (extrusion, rotation, and scraper deflection), and a six-axis coordinated motion control system [109]. The cylinder piston extrudes the concrete material from the nozzle. By controlling the angle of the scraper, the required building shape can be obtained during rotation, and the concrete surface treated by the scraper is relatively smooth and flat. Its rapid, efficient, and economical construction materials have advantages in building low-income housing and emergency shelters. The ability to build complex curves and other geometric shapes can also be applied to building unique buildings.

The large-scale 3D printing of ultra-high performance concrete (Figure 14) developed by XtreeE SAS [110] consists of a system command, robot controller, printing controller, robotic ARM, printhead, accelerating agent, and peristaltic pumps for the accelerating agent and for the premix mixer. Because it does not have a scraper for contour machining, a smaller nozzle is used to obtain a better surface. Printed product surfaces will produce a unique rib finish, and many industrial arts printed with concrete use this feature to show artistic effects. However, if a smooth surface is needed, the wet material is wiped during construction, or the printed surface is ground to a smooth surface. Since there is no other external force applied to the concrete extruded from the nozzle to the print, there will be small gaps between the layers in the printing process, which will affect the strength of the print. High-strength concrete is developed to compensate for the strong influence caused by the existence of this pore structure.

The D-shape printing machine (Figure 15) developed by Monolite Ltd. includes an XYZ gantry system [111], an adhesive distribution system, and a series of nozzles arranged on the gantry. The principle is that a layer of material with the required thickness is laid in the largest printing area and compacted. Then, the nozzle installed at the specified position on the gantry releases the deposition adhesive to make the structure of the desired part solid. Thus, the next layer of material is added to continue the operation, the unbonded material is removed, and the printed part is presented. Since this process is solidified layer by layer, it is not necessary to consider the problem of adding supporting parts to the suspended position.

Whether it is a gantry-type or manipulator-type concrete 3D printing machine, its motion range is limited. So, the final printed architectural structure size cannot fully meet the demand. Furthermore, different buildings need different structures and materials because of the particularity of buildings, which makes it difficult to standardize the products or easy to computerize. Thus, production on a large scale is not easy.

Using autonomous robotic systems to virtually monitor the work-in-progress during the whole construction process is a flexible approach to improving safety, quality, and productivity in construction engineering and management [112]. The rapid advances in sensing, 3D reconstruction, and artificial intelligence, along with autonomous navigation algorithms and low-cost cameras, have made autonomous robots more affordable, reliable, and easy to operate for construction-used surveillance [113]. Generally, these robots include two categories: (1) unmanned aerial vehicles (UAVs), namely, drones; (2) unmanned ground vehicles. By capturing very large collections of images and videos and processing these visual data into 3D models by computer methods, construction-used surveillance robots can frequently serve in the different phases of the whole construction process, such as land surveying, logistics, onsite construction, maintenance, and demolition.

As a key procedure at the beginning of the construction process, land surveying is a fundamental part of the construction industry. Conventional land surveying methods need bulky tools, such as tripods, GSP waypoints, and total stations, to provide navigation and data collection capabilities. Drone-based surveillance robots equipped with cameras, autopilots, and image-processing software can be applied to provide faster and less costly land surveying and mapping in constructions [114,115]. One study compared these two measurements, a manual ground-based survey and a photogrammetric survey, by using a drone-based surveillance robot [114]. A camera mounted on the drone is pointed down towards the ground to measure 3D dimensional coordinates, and 3D models can be then created by aerial photogrammetry (Figure 16). Compared with a traditional GPS survey, the drone-based surveillance robot can reduce the time to one-third and greatly increase the point density by more than 3000 times through the process of autonomously collecting color-coded points. In addition, Fleming et al. [115] used a smartphone-controlled drone to measure the geometry of the support system during the excavation stage of a construction project. Two-dimensional images were converted into three-dimensional construction models to guide different types of construction activities, e.g., site geometry change, geotechnical engineering evaluation, and responses of adjacent structures. Using this method, the construction activities of the project can be adjusted in time.

Autonomous surveillance robots are also effective tools in onsite constructions (Figure 17) [112]. The visual performance monitoring procedure using this method is to first collect onsite images or videos from the most informative views; secondly, analyze them about performance deviations, especially for progress and quality during construction; thirdly, monitor ongoing operations for productivity and safety; then characterize the as-is conditions of existing construction sites; finally, visualize and communicate the most updated state of work-in-progress. Most of the studies use GPS-based drones and predetermined flight paths to achieve autonomous navigation and data collection [116,117,118]. However, these drones for onsite construction surveillance are negatively affected by poor GPS signals in indoor environments or by shadows caused by nearby buildings or other structures. They can also be subject to navigational hazards due to the loss of calibration on magnetometer sensors close to steel components. Therefore, recent efforts have been devoted to Simultaneous Localization and Mapping (SLAM) techniques [119] for both drones and unmanned ground robots. Kim et al. [120] developed a drone-assisted automated framework for autonomous ground robots equipped with laser scanning systems outdoors with the aid of drones. They use a coarse 3D point cloud generated by a drone as an initial map of the target site to determine the scan location and path (Figure 18). A mobile ground robot can collect data at stationary laser scan positions with less occluded views while capturing crucial geometric information as much as it can. Optimal paths for the autonomous robot to navigate among the estimated scan positions are finally generated.

With a hugely increased amount of aging buildings, inspection and monitoring are of great importance to the existing civil structures. Their lifespans can be extended by evaluating their structural states. Conventional methods normally require special equipment and specially trained personnel involved in the inspection process, which can lead to high costs. However, drone-based autonomous robots only require an operator on the ground to control the flight path and camera monitoring the drone. Therefore, they have been exploited in visual inspection and damage detection of construction structures, e.g., building damage [121]. With a remote sensing system mounted on drones, high-resolution images can be obtained from decimeters to centimeter scale (Figure 19) [122]. Capturing images at appropriate scales allows us to autonomously perform comprehensive damage assessment by determining different levels of damage evidence from complete collapse to cracks in buildings. Drone-assisted surveillance robots have also demonstrated the potential to identify and report defects on tunnel lining surfaces (e.g., water leakage, spalling, and cracks) [123].

Among the above applications, autonomous driving technology has revolutionized the more tractable use of concrete construction robots, namely autonomous mobile robots [124,125,126,127]. Unlike cars functioning on city streets, autonomous mobile robots in construction are at a point where they can be safely integrated by utilizing the same equipment as self-driving cars. However, robot navigation to achieve autonomy is a very challenging problem, especially in construction sites. The unknown environment requires the ability of safe path planning in construction robots, particularly for obstacle avoidance of autonomous control. Therefore, Zohaib et al. [128] developed an intelligent obstacle avoidance algorithm, called Intelligent Bug Algorithm (IBA), to navigate an autonomous mobile robot. This algorithm can well accomplish the goal in relatively less time compared with other algorithms (Figure 20). Iqbal et al. [129] presented a specialized tracking system for a tethered-based robotic vehicle, RObot for Scientific Applications (ROSA), which could also be also used in other tethered robots for construction after minor modifications (Figure 21). The proposed system is capable of tracking the tether within a resolution of ±6 cm for the rover’s 40 m. After solving the above problem, autonomous mobile robots can be applied to various construction processes. For instance, an autonomous mobile robot named Romu was developed for working with small-scale sheet piles (Figure 22) [130]. Sheet piles can be carried into a target setting and driven into the ground sequentially, creating a sturdy wall that could serve as a check dam to assist in reducing flash floods in the arid environment. When inserting sheet piles into the soil, Romu uses a vibratory hammer and relies on its own weight to drive piles deeper without carrying excessive additional mass.

The study of concrete construction robots is a new field that combines construction and the research of robots. Its wide application range can cover the whole duration of construction. Challenges and opportunities in concrete construction robots for long-term development and improvement, together with that of autonomous robots in concrete construction, will be discussed in the following directions.

Researchers and engineers have made significant progress in robotics and automation for construction over the last decade. Driverless construction equipment, robotics, and electric vehicles (aerial or ground) are considered ideal solutions to environmental construction problems and labor shortages. However, the main difficulty of them is teaching the machines to deal with the complexities of construction sites. Generally, construction robots can be categorized into three types: remote-controlled, semi-autonomous, and fully autonomous. Existing concrete construction robots need more workers or technicians to participate in the construction process, and workers need to deal with technical problems encountered during operation. Most of them belong to the first two types, and high-level automation has not yet been fully achieved. The future concrete construction robot should be equipped with multiple types of sensors, combined with large data technology, independent analysis, and processing of construction problems encountered so that a fully autonomous level can be achieved.

Most of the existing concrete construction robots are large and require very large construction space. Some details, such as corners, cannot be handled well and are not convenient for assembly, disassembly, and transportation. The future trend in construction robots is they are designed to be miniaturized, light, and efficient. This can be achieved by improving battery capacity and innovating navigation technologies and devices, such as the transforming Tesla drone concept. An advanced lithium-ion battery allows the drone to extend flight time by up to one hour on a single charge. In addition, construction robots equipped with cameras can be controlled by tracking the head movements of virtual reality (VR) headsets. The combination of VR and wearable devices can fully visualize the construction sites and building components, which further reduces some physical equipment or sensors on the robots. After the concrete construction robot is miniaturized, the movement is subsequently more flexible. Thus, narrow spaces can be handled, and transportation at the end of the project can be facilitated.

Digitization in construction concentrates on the centralized building information model, which reforms the construction industry by embracing mature technologies. The traditional concrete construction robot sensor accuracy is low, there will be a small deviation, and the robot will not be able to accurately locate and provide correct feedback due largely to the limitation of GPS, resulting in a poor construction effect. However, the use of high-precision sensors will increase the design costs, contrary to the original intention of reducing costs. Recent studies have been exploring innovative navigation systems and algorithms for robotics, meaning that future construction robots will be able to dispatch from the reliance on GPS satellites. Until then, the construction robots can autonomously move inside the structures, infrastructures, and buildings under construction, such as that underground, in deep canyons, and in other places with poor or unavailable GPS signals. Combining the sensor data with the advanced navigation systems and algorithms, the high-precision construction effect can be obtained by optimizing the motion trajectory and posture. The new technology will achieve more comprehensive construction quality inspections and time management.

Even for extremely simple residential construction, plenty of skilled workers are required for repetitive tasks, which need to be not only successfully completed but also effectively coordinated with each other. Task coordination has been demonstrated in industrial fields with homogeneously distributed robots, heterogeneous robots, and multi-system integration of drones and ground vehicles. For instance, by simply following pre-planned paths without communicating with each other, autonomous excavators and haulage trucks can still be operated simultaneously. This level of task coordination is years ahead of the level that can be achieved in construction sites. Existing concrete construction robots are independent, and the connection between their constructions is not smooth, resulting in a waste of time. If the cooperative construction system among concrete construction robots can be established, multiple concrete construction robots can participate in the construction together, completing different work without interfering with each other. It can be monitored in real time, achieve partial overlap of each process, and shorten the construction period. For full autonomy in construction, advanced autonomous systems should be developed to achieve inter-robot coordination among robots performing various tasks.

Autonomous driving and monitoring technologies have been gradually proposed and applied in concrete construction in terms of leveling and compaction robots, floor finishing robots, surface painting robots, 3D printed construction robots and construction-used surveillance robots. However, some shortcomings or challenges of using autonomous robots should be addressed in the construction process. First, the use of these robots, especially drones, can be limited by local regulations, which vary from region to region. Second, using autonomous robots in concrete construction requires professional operators, as controlling these devices can be challenging. Third, autonomous robots on construction sites face a critical issue of moving reliability since rain and strong winds can negatively affect moving paths. To ensure reliability and enhance applicability, more reliable algorithms that can cope with a variety of conditions must be developed. Fourth, there is still a lack of electricity and payload capacities with the robots, which limits their operating time. Fifth, equipping more devices on the robots requires a better power-generation module. Last, the automated level of these technologies needs to be improved because many of them need to be accompanied by manual detection by construction workers.

In the modern construction industry, autonomous robots are one of the most innovative technologies which have the promising potential to facilitate various construction activities. These activities can involve not only concrete construction but also different phases of tasks over the whole construction process, such as construction sites survey, onsite constructions, work-in-progress monitoring, document records for progress and safety, constructed structures inspection, leading to savings in time, costs and injuries while improving construction quality.

Companies and institutional researchers have observed the valuable benefits of using autonomous robots in construction work. First, injury and fatality can be greatly reduced or avoided through the effective use of robots. Having a mapping solution based on autonomous robots yields automated movements or flights to reduce risks associated with surveying construction sites, such as heavy equipment and hazard-related injuries. Autonomous-robot-based technologies can also resolve the difficulties of dangerous and hard-to-reach structural inspections, including but not limited to steep-sloped roofs, walls, and facades on the exterior, towers, and bridges, damage from complete collapse to cracks in buildings. Second, autonomous robotic approaches are relatively cost-effective and can improve production efficiency. Rather than using human resources, heavy equipment, and expensive tools, drones can achieve rapid data collection and automatic analysis with impressive accuracy in construction site surveying, monitoring, and inspection, significantly reducing project costs. Autonomous robots can also avoid the high personnel costs that repetitive works of onsite constructions require, such as concrete distribution, leveling, compaction, floor finishing, and surface painting, while leading to a higher level of productivity at the same time. Last, unmanned aerial and ground vehicles in construction using autonomous robotic technologies have been determined as ideal solutions to environmental construction problems and labor shortages. In the future construction industry, drones could be used to guide autonomous robots on the ground in safe and efficient patterns. Consequently, autonomous robots collaborating with drones on construction sites will be able to reduce fuel consumption, shorten construction schedules, and improve the safety of construction processes. In conclusion, we believe that developing autonomous robots in construction will be a trend.

As an emerging technology with great development potential, concrete construction robots are expected to realize “safer, more efficient, greener and more intelligent” construction, and the whole construction industry must take this opportunity to complete this leapfrog in development. In this paper, the technical progress of concrete construction robots and autonomous robots in concrete construction are described systematically. Meanwhile, some development suggestions are put forward from several perspectives.

This review provides a new perspective based on the basic functions during the concrete construction process, and the mechatronic system of autonomous construction robots can be developed accordingly. This new insight builds a close connection between the basic demands of highly repetitive or dangerous construction tasks and the functional structures of autonomous construction robots. We divide the existing concrete construction robots both on the market and under study into six categories: distribution robots, leveling and compaction robots, floor finishing robots, surface painting robots, 3D printing robots for additive manufacturing, and construction-used surveillance robots. The general characteristics, functions, and applications of the robots are described and compared systematically, which could guide developers to better design robots that closely match real construction needs. The mechatronic system of an intelligent construction robot is supposed to be achieved by integrating functional mechanical parts, autonomous control systems, and highly precise sensors. It is hoped that this paper will provide inspiration for future researchers to grasp the future trends of robotic research in concrete construction.