Doosan Bobcat announced on November 16 that it has successfully completed a field demonstration of autonomous construction equipment operation technology with AI solutions company Maum.AI.
The two companies signed a memorandum of understanding (MOU) in early June and have since been validating the feasibility of autonomous operation technology. As a result, in just about five months, they implemented technology that enables compact construction equipment to work autonomously in a real work environment.
The demonstration utilized Vision-Language-Action (VLA)-based AI technology. It perceives the surrounding environment through sensors such as cameras, understands the work situation, and then translates this into the equipment's actual movements. Unlike conventional automation that moves according to preset rules or human commands, it is characterized by recognizing the work environment and situation, making judgments, and acting accordingly.
Doosan Bobcat selected manure handling in livestock barns as its first demonstration task. A Doosan Bobcat representative explained the choice, saying, “Livestock farms are suffering from a chronic labor shortage due to rural depopulation and aging,” and “Among their tasks, manure handling is one that many people avoid because of the strong odor and heavy workload.”
▲Doosan Bobcat's skid steer loader scoops manure into its bucket and moves it into a trailer without operator intervention.
The final demonstration, held at a cattle barn in Pyeongtaek, Gyeonggi Province, was broadcast live to a meeting room at Bundang Doosan Tower, where Doosan Bobcat executives watched the work process. In the first demonstration, the Doosan Bobcat skid-steer loader autonomously and repeatedly performed a series of tasks: △scooping manure into the bucket, △moving it to a designated location, and △dumping it into a 2.4m-high trailer.
▲Doosan Bobcat's skid steer loader detects and cleans up manure scattered across the floor without any operator intervention.
In the second demonstration, it △recognized manure spread on the floor, △independently determined a cleaning path, △moved forward with the bucket kept as close to the ground as possible to collect the manure, and △piled it at a designated location, repeating the task several times in succession. Both demonstrations were completed successfully without operator intervention.
Charlie Park, Executive Vice President and Chief Strategy Officer (CSO) of Doosan Bobcat, said, “This demonstration is the first fruit of the technology validation the two companies have carried out together since June,” and added, “We will expand collaboration among our global R&D teams and with diverse partners in Korea and abroad to advance autonomous operation and unmanned technologies, and develop them into practical solutions that improve customers' productivity and work efficiency.”
Chris Choi, Chief Technology Officer of MAUM.AI, said, “By optimizing MAUM.AI’s proprietary Physical AI model for livestock-barn conditions, we achieved a precise construction equipment control and enhanced autonomous performance.” He added, “The successful demonstration of autonomous manure handling—a labor-intensive task that workers often avoid—shows Physical AI’s potential to solve real-world challenges across industrial worksites.”
Doosan Bobcat plans to develop autonomous operation scenarios for simple, repetitive work and for jobs that are difficult or burdensome for people, across industrial sites such as livestock, agriculture, landscaping, and construction, and to broaden their scope of application.