Robotics is now more promising direction is no more than: industrial robots, unmanned aerial vehicles (UAV), unmanned vehicles (UGV), medical robots. The industrial robots are now in increasing demand, is a good industry, but at present the main research are companies, the university almost do not do (but the university will use the industrial robots to do some other research, such as grasping research, that is, the face of irregular objects, how to determine the shape, how to use the robotic hand to accurately and stably grasping).UAV is mainly the government is investing in the money, the military use.UGV is the government and some of the forward-looking companies. UGV is used by the government and some forward-looking companies, and I'm afraid it can't be commercialized on a large scale within ten years according to the current technology. Medical robots are aimed at some of the current surgical problems, there is some progress (da Vinci system), the commercial outlook is very good, but I'm afraid that in the short term, the promotion is not open. Other directions, such as domestic robots, Willow Garage is exploring, but the prospects are not optimistic. Japan's humanoid walking robots are doing very well, but the application prospect is extremely narrow (at present, they can only be used in disaster environments where it is difficult for human beings to penetrate to operate facilities originally designed for human beings).Boston Dynamics's quadrupedal walking robots are doing well, but the scope of application is too narrow, and only the military may be interested in purchasing them.Sarcos Raytheon's exoskeletons are doing well, and there is a certain commercialization prospect, but the weight is too heavy, and it is too heavy for the army. It has some prospects for commercialization, but it weighs too much, has high energy requirements, and is very costly.
There are three main levels of doing robotics.
The bottom level does mechanical design, mainly motors, sensors, artificial muscles, structural design and so on.
The middle level does control design, mainly sensor filtering and control algorithms.
The upper layer does perception, mainly based on sensor data for higher-order information fusion, for more complex analysis, such as machine vision, SLAM and so on.
Artificial intelligence in the field of robotics currently do not see much use. Even machine learning is used very sparingly. All three levels are good. But it's the last two levels that are currently the fastest growing. The middle level is growing faster in industry right now. The upper tier is still mostly in academia, and there's not much to commercialize yet. The lower level is very slow right now, mainly because the very best materials have not been found.