Spanish Researchers Reveal Dual-Arm Robot That Learns Housework by Watching Humans

Researchers at the Carlos III University of Madrid, known as UC3M, have introduced a significant development in domestic automation through the ADAM robot. The Autonomous Domestic Ambidextrous Manipulator, or ADAM, is designed to perform complex household chores by observing and imitating human actions, representing a shift from rigid programming to flexible learning in the field of service robotics.

The research, recently highlighted by the Mobile Robots Group at UC3M, focuses on the technical challenge of bimanual coordination. While single-arm robotic systems are common in industrial and some domestic settings, the ability to coordinate two arms simultaneously to handle fragile objects or perform sequential tasks has historically been a bottleneck for engineers. The ADAM platform addresses this by teaching each arm to perform tasks independently through imitation learning, before allowing the system to synchronize these movements autonomously.

Measuring approximately 160 centimeters in height and equipped with a mobile base, the robot uses a combination of 2D and 3D laser sensors alongside RGB cameras with depth information. These tools allow the machine to generate three-dimensional models of its environment, which is essential for navigating narrow domestic spaces such as doorways and kitchens. The system is currently capable of performing tasks such as setting and clearing a dining table, tidying rooms, and delivering medication or water to users at specific times.

Director of the Mobile Robots Group, Ramon Barber, stated that the robot is specifically intended to support elderly populations or those with limited mobility. By taking over repetitive or physically demanding daily chores, the technology aims to allow individuals to remain in their homes longer while reducing the physical burden on human caregivers. The robot is built to respect personal space and operate collaboratively, making it suitable for residential environments rather than just clinical or industrial facilities.

The technical framework of the ADAM robot is divided into three distinct phases: perception, reasoning, and action. During the perception phase, the robot identifies and locates objects within a room. The reasoning phase involves processing this data to decide on the best course of action, while the action phase executes the physical movement of the mobile base and the dual arms. This process ensures that the robot can adapt to unexpected changes in its surroundings, such as a moved chair or a person walking into its path.

While the ADAM robot is currently functioning as an experimental platform, the development team has indicated that the underlying technology is maturing rapidly. Estimates suggest that while the current prototype costs between 80,000 and 100,000 euros, advancements in manufacturing and sensor technology could make these units commercially viable for home use within the next 10 to 15 years.

In the context of the broader construction and smart-home industry, the integration of such autonomous systems is expected to influence future residential design. Architects and developers may need to consider the spatial requirements and charging infrastructure for mobile assistive robots as they become standard features in modern living environments. For now, the UC3M project serves as a technical benchmark for how imitation learning can bridge the gap between human labor and robotic assistance in the domestic sphere.