Artificial Intelligence in Material HandlingResearch into AI is well funded and progressing quickly in the material handling arena. As reported by the NY Times, researchers in Berkley, for instance, are working on robots that can look inside a bin full of randomly sized objects, quickly understand the size and shape of each discrete item, and then successfully grab one and place it in a desired location. This type of behavior has always been outside the capability of robots, requiring companies to hire human “pickers” to perform it. Other areas of research include self-navigating autonomous industrial vehicles (e.g. AGVs and fork trucks), using AI to optimize material flow in a facility, predictive maintenance for material handling components, and many more.
There are also many AI applications that have already moved from research to employment in facilities around the world. Self-navigating industrial vehicles, although still an area of research, are already in many warehouses (continued research will serve to increase their capabilities). Omron’s mobile robot, Adept, for instance, will automatically drive to desired locations once it has mapped a facility. With the ability to grab objects of different shapes, there are even some robotic grippers that are already employing a portion of the technology being researched by the Berkley group. Clearly, then, AI has had a substantial effect on material handling automation, and with continued research, it will continue to shape it. Does this mean we can expect a future where distribution centers and manufacturing facilities are totally automated, i.e. without any human workers? Perhaps, but probably not.
To date, AI has increased the capabilities of material handling automation, and automation, in turn, has made workers more productive and valuable. Thus, in my experience, after automating a process, companies do not eliminate workers. Rather, those workers become much more productive or are moved into a different process that provides more value. For example, a goods-to-person system such as AutoStore will drastically increase the number of items a human picker can pick, so the human is not replaced, but is now much more productive. Eventually, I do think that researchers like those at Berkley will create systems that replace the human picker in this process, but that person will not go away. Machines are not perfect, and humans will need to supervise them and correct them when they make mistakes. The picker, therefore, will become a supervisor of intelligent machines, taking immediate corrective action when the machines make mistakes in order to keep production moving.