Robotics had traditionally required very detailed programming in order to get meaningful work out of a Robot. Major advances are being made in the area of Robot Learning, both in the performance of repetitive tasks, and in the area of responding to natural language commands.
In industry, previously heavily capitalized operations could afford to purchase and program Robots. Today, cost effective robots that can be easily programmed-by-example to do repetitive tasks in small to mid-sized businesses. The “Baxter” Robot can be purchased for $22,000 and easily programmed to do a wide variety of manual tasks by ordinary workers. Here is Baxter solving Rubik’s Cube.
More ambitious are projects that allow Robots to learn and be productive in unstructured Human environments. Projects such as Robo Earth and TellMeDave represent a new type of crowdsourcing of robot learning that promises to jump-start worldwide Robot learning and collaboration. Here is the best presentation I have seen as to how this type of learning takes place. This is a real eye-opener into how Robots will move abut our world performing tasks based upon a deep pool of common Robot understanding.