The challenge of recruiting construction workers is increasing the urgency of discussions about how heavy machinery should operate in the future. According to the 2025 AGC and NCCER workforce survey, 92% of construction firms looking to hire reported difficulties finding qualified workers. For companies exploring automation, this creates an important practical question: how much can software achieve when the machinery it controls was originally designed for a human operator?
The same survey found that labor shortages had contributed to project delays at 45% of construction firms. These figures provide a clear reason to explore alternative approaches to organizing work. However, they do not establish that AI will eliminate workforce shortages or that existing equipment is automatically ready for autonomous operation.
Research in connected manufacturing provides a useful starting point. The National Institute of Standards and Technology explains that digital twins, which are computer-based representations of physical systems, can help with monitoring, forecasting, and maintenance planning. Developing these models requires collecting relevant data, validating their accuracy, and connecting information from the systems they represent.
When applied to heavy machinery, this approach raises an important question before autonomy can be considered: can software obtain a reliable understanding of a machine’s condition and behavior? A digital model is only as useful as the quality of the information it receives and its ability to support sound decisions.
Hiten Sonpal, CEO of RISE Robotics, a company developing electric alternatives to hydraulic motion systems, believes this connection deserves more attention. He argues that conversations about industrial AI can focus too heavily on software capabilities without adequately considering whether the underlying machinery can respond consistently and predictably.
“The companies that are making these AI systems move at a different pace than the companies that actually design the heavy machines themselves,” Sonpal says. In his view, this gap can encourage businesses to add intelligence to existing equipment, even when more fundamental engineering changes may be required.
His concerns focus on the relationship between a digital command and the machine’s physical response. Sonpal explains that pressure fluctuations and flexibility within hydraulic components can make precise movement more difficult to achieve. He presents this as an engineering problem that requires accurate control and continuous feedback, rather than proof that all hydraulic machinery is inherently unsuitable for automation.
Introducing monitoring systems can create additional challenges. Cameras, sensors, and electrical wiring must operate reliably in demanding work environments. Sonpal believes manufacturers need to consider how these components work together and whether the complete system provides software with enough reliable information to make effective decisions.
“Not only do you need precise control to move everything, but you also need precise sensing to understand what’s actually happening,” he says. Measurements of position and force can help determine whether a machine’s actual movements correspond with its intended actions.
For this reason, Sonpal supports incorporating electronic controls and feedback systems into machinery from the beginning. He views drive-by-wire technology, which uses electronic commands to control physical movement, as an important foundation for remote operation and more advanced automation, including applications such as Beltdraulic. Accurate sensing, he adds, can also improve the usefulness of a digital twin by ensuring its representation remains connected to the machine’s real-world activity.
Human operators will continue to play an important role in this approach. “Even before you can get to AI, you have to find a way to let human operators drive these machines remotely and step in when needed,” Sonpal says. He expects unusual situations to continue demanding human judgment, making the ability to intervene an essential part of equipment design.
Maintenance creates another challenge that automation alone cannot resolve. Controlling a machine’s movements does not automatically determine who will inspect its components, detect leaks, or coordinate necessary repairs. “It’s not just moving the machines. It’s about making sure that they’re operational with minimal assistance,” Sonpal says. He believes these responsibilities must be addressed alongside the technology used to control the equipment.
Looking toward the future, Sonpal envisions construction sites where drones map working areas, software coordinates machinery, and human operators step in when equipment encounters unexpected situations. He presents this as a potential model for future operations rather than an established outcome. He also points to noise, vibration, and other workplace conditions as factors worth considering when evaluating recruitment challenges and the possible benefits of remote working arrangements.
For equipment manufacturers and buyers, Hiten Sonpal recommends starting with a practical assessment of the machinery itself. Companies should determine what their equipment can control, what it can measure, and how human operators will intervene when problems arise. He argues that these considerations should guide investment decisions before ambitious expectations about autonomy turn into firm commitments.
The broader message is that industrial AI requires a reliable physical foundation. Meaningful progress will depend on designing machinery, sensing systems, maintenance processes, and human oversight as connected elements, ensuring that each supports the outcomes the equipment is expected to achieve.
What Vehicle Tracking Data Reveals About Operational Bottlenecks
