Hadrian looks like a truck crane, but he builds like a world champion. The star of the Australian company FBR manages 200 extra-large blocks of aerated concrete or clay per hour. This corresponds to 2200 manually laid standard bricks. Hadrian just pulled up the walls of an entire daycare center in Perth in 57 hours. Even the best bricklayers cannot achieve such values. And the construction robot should be much faster: a model with a laying speed of 1000 blocks per hour is in development. Hadrian reads the correct position of the bricks from stored 3D construction plans. He not only layers the elements, but also cuts them to the required length if necessary – without human intervention, but with the greatest precision.
Automation in construction will experience a boom, says Sami Atiya. “Anyone who walks through a construction site with trained eyes sees the enormous potential for efficiency.” The robotics boss of the global technology company ABB doesn’t just mean the time and cost savings. According to a survey by the group, 91 percent of 1900 construction companies surveyed in Europe, the USA and China expect a shortage of skilled workers in the next ten years. Robots could fill that void. Above all, repetitive, dangerous or unhealthy work could be carried out by autonomous, i.e. independently acting, machines. For example, demolition work that releases asbestos or lifting and transporting cement sacks.
The problem: Every construction site is different and, moreover, is constantly changing
To do this, however, the artificial workers have to move freely and find their way around the construction site. This is a challenge for robots, which up to now have mostly monotonously carried out the same work steps in a controlled factory environment in mass production. Because every construction site is different and on top of that changes every day. Sometimes the ground is dry, sometimes it is muddy after a downpour. Where there was soil yesterday, there is a hole for a shaft today. In addition, there are always deviations between planning and construction site reality. It is enough if a door is installed offset by half a meter to literally let the robot run against the wall.
But robots are getting smarter and smarter. Sensors, scanners, cameras and image processing programs help you to find your way around even in changeable surroundings and to master new, demanding tasks.
In autumn 2020, the tool manufacturer Hilti presented Jaibot, its semi-autonomous construction site robot for drilling ceilings, for the first time. Although it still has to be navigated through the construction site by remote control, the robot identifies the drilling jobs within its range on site, executes them automatically and marks them in color for the various trades.
Robots could help with overhead work in particular
In particular, it relieves the craftsmen in the heating, air conditioning and ventilation sectors with repetitive, physically strenuous work. “We looked at which routine work on the construction site is one of the most stressful, and that is primarily the overhead work,” says Julia Zanona, Product Manager for Robotics at Hilti. In order to be able to carry out its tasks accurately, i.e. to drill the holes in the right places in the required size and depth, Jaibot uses special data.
With Building Information Modeling, or BIM for short, all data of a building – work plans, the chronological sequence of the work of the various trades, plan changes, information about materials, maintenance dates – fed into a digital database, networked and constantly updated. The information relates to the entire life cycle of a structure from design through construction to operation. All project partners can access it in the cloud. That means: everyone involved in the construction and operation of the building has an insight into all work steps and processes at all times. Everyone is always up to date – including Jaibot.
However, this knowledge alone is of no use to the robot; it has to connect it to its control system in order to be able to act autonomously. A team of researchers at the Fraunhofer Italia Innovation Engineering Center in Bolzano has developed such a software interface. ROSBIM connects BIM with the “Robot Operating System”, or ROS for short. The aim of the researchers is to make optimal use of building data that is already available digitally on the construction site. For example, for the transport of heavy loads such as building material or tools with an autonomous, mobile robot.
Husky A200 is designed for harsh environments
In Bolzano, the team is training with the commercially available Husky A200 robot carrier platform. It is designed for harsh environments, rolls on wide tread tires and is equipped, among other things, with laser and inclination sensors that help it to navigate in rough terrain. However, Husky still has a lot to learn.
“A person can react intuitively to sudden changes in their environment, the robot has to access its sensor data,” explains Michael Terzer, research associate on the ROSBIM project. As an example, he cites a hole in the ground that many environmental scan sensors could not detect. It is stored in the data that the hole, for example an elevator shaft, is open on a certain day for maintenance purposes and must be bypassed. “Such time-dependent data, which a robot cannot detect with the help of its sensors, is received via the interface,” says Terzer.
Together with colleagues from the Fraunhofer Institute for Industrial Engineering (IAO) in Stuttgart, the Bozen-based Fraunhofer team developed Balto, an autonomous disinfection robot named after a sled dog that brought urgently needed vaccine serums to an area of Alaska that was difficult to access a hundred years ago. Balto is also equipped with ROSBIM and can therefore react to its surroundings. For example, in order to thoroughly disinfect all door handles in a hospital, he has to know where and at what height they are, what material they are made of and how often they are used. The robot can read this information from the data and select the optimal disinfectant.
“You don’t need to feed Balto with coordinates, you can commission the disinfection of a whole class of objects,” says Günter Wenzel, head of department at Fraunhofer IAO. Instead of having to tell the robot the location of each door handle to be disinfected individually, a general order such as: All heavily used door handles should be disinfected with a frequency of X minutes. Balto also plans his disinfection route himself. Because of the transmitted data, he knows in which room and at what time a meeting is taking place.
Who is actually liable if something goes wrong?
According to the researchers, the ROSBIM interface is also important for human-robot interaction. If Balto encounters one or more people, he should evade. If this is not possible, he stops. However, it would be bad if he stopped in an escape route of all places and blocked it. However, since the course of the escape routes is also recorded in the BIM data, such situations can be avoided.
The interface can not only be used on one side, but in two directions. The robot can report its work progress to the system and coordinate disinfection tasks with other Balto colleagues.
Balto is by no means limited to his current area of responsibility, but can also combine disinfection with cleaning and, in the long term, even take on tasks in the field of construction site monitoring and building maintenance. For the time being, the prototype is only working in the NOI technology park in Bolzano, monitored by the research team. Further demonstrators are in use in the Future Work Lab at Fraunhofer IAO and in the Center for Virtual Engineering.
The question arises: Will autonomous robots replace the worker on the construction site, the cleaning worker or the maintenance technician in the foreseeable future? “Do not replace, but support,” says Michael Terzer. “In order to be able to use the technology on a large scale, it still needs a lot of further development and that will take time.” Aspects of liability also have to be clarified first. So it remains exciting.