Researchers Develop Machine-learning Technique That Can Efficiently Learn To Control A Robot

Researchers from MIT and Stanford University have developed a new machine-learning approach that can be used to control robots in dynamic environments. This technique could help an autonomous vehicle learn to compensate for slippery road conditions, enable a drone to follow a downhill skier despite strong winds, or allow a robotic free-flyer to tow different objects in space. The approach incorporates structure from control theory into the process for learning a model, which leads to an effective method of controlling complex dynamics. The technique is able to learn an effective controller using fewer data than other approaches, which could help the system achieve better performance faster in real-world applications.