RUB arrow Bauwesen arrow Informatik arrow
rviz pointcloud

Unmanned Aerial Vehicles in Construction

DSC00096The emergence of 4D BIM in construction and the resulting need for feedback data about the state and progress of construction processes raises the question of how to implement efficient and rich monitoring solutions. One particular aspect is the visual inspection of outward-facing structural elements, especially with high rise construction projects. Traditional methods of documentation, inspection and progress-monitoring cannot fulfill the new requirements towards richness of information and high update rates.

Employing UAVs with modern autonomous capabilities as a tool for visual inspection tasks offers substantial advantages over traditional inspection methods.

  • The ability to reach remote vantage points in great heights or otherwise inaccessible locations
  • Quick deployment and little to zero interference with ongoing processes
  • Accurate positioning offers structured and repeatable inspections that yield comparable results over time

Our research aims at the development of an automated, BIM-driven workflow for visual inspections in construction.

Autonomous Robotics Technology

The design of an automated data generation concept implies modern robotics concepts, based on open source autopilot software (PX4) and the Robot Operating System (ROS). A central concept of this proven approach is a simulation driven development, in which a robot behaves identically in simulation and real execution.

Realistic simulation with real robot behaviour enables quick development with meaningful insights. Autonomous robots are developed in simulated environments with physics engines. Execution on the actual robot is analogous as it only requires the sensory input and actuators to be switched. This concept enables rapid development and testing without the risk of damaging hardware.

After investigating GPS-driven navigation on construction sites, focus shifted to reactive autonomy which includes the UAV’s ability to sense the physical environment and actively avoid contact with objects around it. Point clouds derived from stereo cameras provide reliable information for navigation and furthermore provide meaningful representations of the as-built status.

An Information Loop Designed around BIM

BIM data plays a fundamental role in the autonomous navigation concept. As-planned information serves as a description of the known environment before takeoff. Knowledge about the environment is necessary for finding optimal flight paths to target structural elements. Without knowledge about its environment the agent would be forced to explore its environment each time it receives a new target location.

Closing the loop, the information gathered by the UAV is fed back into 4D BIM. Each inspection flight creates a new instance of as-built data, which also serves as an updated known environment for subsequent flights.

Furthermore, differences between as-planned and as-built status can be measured and visualised in real-time.


Project Data

Title: Unmanned Aerial Systems in Civil Engineering


Researcher:Henk Freimuth

Chair of Computing in Engineering
Bldg. IC, Room 6-59  
Universitätsstraße 150 
44780 Bochum