If robots are going to be useful for everyday tasks, they need to combine skills, adapt to changes and complete complex, multi-step tasks. Researchers in the Figueroa Robotics Lab developed SymSkill, which helps robots learn and recombine skills to complete longer tasks. With just five minutes of video demonstrations, a real robot completed sequences of up to 12 steps and could adapt to interruptions and failures. “What makes this exciting is that the robot doesn’t have to experience every possible version of a task before it can perform it,” says doctoral student Yifei Simon Shao. Read the full story. bit.ly/4xT5n6U @MEAM_Penn

The next episode of “Innovation & Impact” features Byron Cook, Vice President and Distinguished Scientist of the Automated Reasoning Group at @awscloud and Professor at University College London. Cook joins host Dean Vijay Kumar to discuss AI, automated reasoning and the evolving relationship between people and intelligent machines. Episode 13 drops next week.