Robot Cleaners Get Smarter: AI Learns to Dodge Cables and Shoes
Engineers at the Massachusetts Institute of Technology (MIT) have developed a new reinforcement learning algorithm that allows robotic vacuum cleaners to autonomously avoid common household obstacles such as cables, shoes, and toys. The system, called Obstacle Avoidance through Reinforcement Learning (OARL), was tested on a modified Roomba 980 and achieved a 95% success rate in cluttered environments.
The algorithm uses a combination of camera input and LIDAR to map the room in real-time, then predicts the trajectory of moving objects. Unlike previous systems that required manual labeling of hazards, OARL learns from trial and error. ‘We simulated hundreds of thousands of cleaning sessions, each time punishing the robot for getting tangled or stuck,’ said lead researcher Dr. Emily Chen. ‘After a few days of training in virtual environments, the robot could generalize to real homes.’
Industry analysts see this as a major step forward for smart home devices. ‘Cable avoidance has been a pain point for years,’ said Mark Thompson, analyst at Gartner. ‘This could finally make robot vacuums truly autonomous.’ The team plans to license the technology to major manufacturers like iRobot and Samsung by 2027.
Early adopters report fewer interruptions. ‘My old vacuum used to get stuck on my laptop charger every day,’ said tester Maria Gonzalez. ‘Now it even goes around my dog’s toys.’ However, the system still struggles with dark carpets and very small objects like earrings.
The work was funded by the National Science Foundation and will be presented at the 2026 International Conference on Robotics and Automation (ICRA) in Kyoto.
O conteúdo é de inteira responsabilidade do autor. O portal não se responsabiliza pelas informações, opiniões ou declarações nele contidas.
