Can't us robots just get along
How industries with collaborative robots may provide us hints on where robots may be going
Recently while listening to the excellent robohub podcast I learned that there have been a lot of advances made in collaborative robots. It got me thinking about what kind systems there might be in real world situations and what kind of real world problems they have had to encounter.
Cooperating robots tend to be rare in the real world. There are enough challenges with deploying one robot in the real world, to deploy two that can cooperate brings with it a whole host of other challenges. If we think about how areas that have a lot of mechanical operators are set up, we see that the manipulators are typically interacting with humans or a fixed site, not other manipulators. A classic example of this is a construction site.
This might be because there are too many variables to coordinate between the machines. The level of sensing and control required to achieve this has to be very high quality even with humans in the loop. Moreover, the gains to this coordination are not high enough to make the level of risk involved in coordinating heavy machinery worth it. This may explain why we have seen the first applications of cooperating robots in areas where the physical size is smaller. With a notable exception being military operations where heavy machinery often interacts with each other in high risk domains, such as airplanes taking off from carrier vessels or airplane dropping tanks from their cargo bays.
At the small scale, there is little consequence of a failure to coordinate and therefore it might make operational sense to attempt this. This is the case the situations covered in this article.
My understanding of medical robots is fairly limited. An oversight that I will try to rectify in the future. One can imagine that in a high consequence environment like a medical procedure, small improvements in operations can have significant consequences. And due to the large amount of healthcare spending, these improvements can also have big economic payoffs. Cooperative robots in the medical sector are primarily seen in the surgical field. Broadly, there are two main types of surgical robots. The first, surgeon extenders, are operated directly by the surgeon and augment or supplement the surgeon’s ability to manipulate surgical instruments in surgery. The second type are auxiliary surgical supports which work side-by-side with the surgeon and perform functions such as holding tools. The reason these robotic type devices are able to collaborate with each other and able to be manipulated by surgeons is that they mostly function as surgical tools. In some ways, it is the equivalent of using a flashlight and a screwdriver at the same time. While there has been a concerted effort to integrate state of the art robotics software and hardware research into these systems, the engineering overhead and certification costs of medical devices is very high. This means that medical robots typically lag in terms of performance compared to robots in other industries.
Another application that is behind some of the design decisions of Boston Dynamics new robot stretch, are warehouse robotics. With the proliferation of warehouse robots such as the autonomous pallet trucks by Vecna robotics, it is easy to imagine a future where robotic arm like robots work with autonomous pallet robots to move boxes around warehouses. These robots would then automate large parts of the dull, dirty and dangerous parts of operating a warehouse. It would also reduce the incidence of the most likely injury resulting from manual materials handling which is muscle injuries. These make up 47% of incidences of injuries in the workplace.
As companies attempt to leverage a variety of robots to tackle tasks they will find that the challenge comes in not interfacing with other robots but with humans to operate those robots safely. Robot to robot communication has been extensively studied in the field of swarm robotics with a variety of methods available to diffuse sensing and control across different platforms. At the same time, there is a lack of understanding on how we can build AI to for robots to collaborate with humans. This was the topic of the earlier sighted robohub episode. A lot of this research is still in very early stages with significant challenges left to tackle. But the amount of progress that has been made in the field of robotics the past few decades leaves me optimistic that over time this problem too will be solved. And we can look forward to a world where we can safely and reliably work alongside robots.