Unmanned Systems Technology Dec/Jan 2020 | Phoenix UAS | Sonar focus | Construction insight | InterGeo 2019 | Supacat ATMP | Adelan fuel cell | Oregon tour | DSEI 2019 | Copperstone Helix | Power management focus

46 Insight | Construction to the system on the amount of dirt it’s moving, and it also calculates the right paths and places to dig, so that it can move freely during the whole process. These have been developed over time through our AI and machine learning models; it’s a continuous loop, with every past project informing future ones.” The Lidar and cameras provide the excavator with perception capabilities to detect obstacles, other vehicles and pedestrians. A GPS antenna is mounted atop the vehicle for gathering geospatial data as well as positioning, and to ensure it stays within a set safety geofence. Several IMUs are also installed across the excavator to give a layer of spatial, motion and balance awareness; a number of additional, proprietary sensors have also been integrated. The company currently has three autonomous construction vehicles, the newest of which is its compact track loader (CTL). That uses the same navigation and sense-and-avoid sensor architecture as the others, but needs less software and hardware because of the smaller arm on it compared to that on the excavator. Developing the autonomy ‘stacks’ powering the vehicles relied on using a mixture of open-source and proprietary technology. Many of the components and connections in their guidance systems have been custom-built to interface with the heavy excavator arms and other vehicle equipment, and to withstand far greater levels of heat, dust and vibration for long periods of continuous operation than most self-driving vehicles would face. Regarding the future, Ahmed says, “We started off with classic construction workhorses – excavators, dozers and CTLs. Now, based on the customers we work with and the requirements of their job sites, we are getting into more specialised equipment, such as trenching machines, new attachments and bucket types, and larger vehicles. Almost any piece of heavy equipment has the potential to become an autonomous robot. “The construction industry – and by extension our entire built world – is facing serious backlogs in building new homes, fixing crumbling infrastructure, creating new energy systems such as solar and wind, and addressing a big labour shortage. By using robots to automate some tasks, we can help tackle these issues and improve the industry and built environment.” Autonomous cranes As improvements in AI enable safe autonomous operations of larger and more complex vehicles, so Israel-based IntSite has developed a sensor and autonomy architecture called AutoSite100 that is aimed at turning construction cranes into autonomous systems. IntSite intends to reduce the number of accidents that typically occur on construction sites, using computer vision for awareness and to ensure the cranes detect and avoid every person and potential points of danger while carrying out tasks. The autonomous cranes are also expected to perform jobs faster and more efficiently than humans, by communicating continuously with each other over RF links to collectively calculate the optimal paths and trajectories for moving materials, while avoiding each other’s arms, loads and pathways. “We’ve already demonstrated prototype systems using our technology to a few construction firms, mainly in Europe,” says Rotem Ravid-Elad, head of product at IntSite. “We’re now just maturing it with an eye to adding more features to further enhance safety and efficiency in the construction yard where we can. “In particular, we’re expanding the range of obstacles for the crane to recognise and avoid, based on how the type of object might move, as well as carrying out avoidance actions faster to reduce the time taken for each task. December/January 2020 | Unmanned Systems Technology Computer vision is key to the operation, navigation and localisation of the crane and its trolley in IntSite’s AutoSite100 (Courtesy of IntSite) Robots allow you to automate some tasks to help tackle serious backlogs in building and address a big labour shortage