Last year I shared Five Myths and Facts about robots, and emphasized their potential in augmenting human labor. Many startups are attempting to leverage the dropping costs of arms, open-source computer vision and deep learning frameworks to replace workers. Unfortunately, most attempts so far resemble “research” more than “product.” In fact, most resemble technology push rather than an outcome of customer demand. There seems to be an emphasis on replacing humans in contrast to augmenting them; which comes with a deep understanding how the robots fit into the operation as a whole. How can roboticists and potential customers work together given the current cost and performance limitations of robotics toward accelerating their adoption in the workplace?
Robots continue to be relatively slow and difficult to program. That hasn’t prevented many startups from finding innovative ways to put robots into commercial and industrial settings, as well as the home. The basic math is simple: If a cheap robot can perform a human task, then the savings from eliminating the human will pay for the robot many times over. Though investors, manufacturers, and logistics companies are putting some money to work towards experimenting, significant capital is yet to flow into robo-startups to make them widespread.
Here is my advice to startups, investors, and potential customers:
STARTUPS: Human workers are not one-trick ponies. Although tasks in logistics and industrial settings are very well-defined, most are more complex than what can be carried out by a robot. People perform many tasks, with the operation designed around people with diverse capabilities. Humans can adapt to changing workplaces on the fly, while robots require (re)programming by experts. Rethink Robotics has introduced the concept of “teaching” robotics with its Baxter. Symbio Robotics has emerged to simplify the process of training robots; however, they are still nowhere nearly as simple as barking instructions to a human worker. Furthermore, many human workers take on a variety of tasks; therefore, the notion of simply replacing the human makes no sense unless the entire workflow is to be revamped — an investment most customers won’t make for an unproven technology.
CUSTOMERS: A robot as a tool vs. replacement for a human. Customers tend to push robotics startups into a corner by asking for a robot’s value proposition in the form of ROI — probably because they have historically made purchasing decisions for everything from forklifts to conveyors on the basis of ROI, while their human operators are not. Robots, while augmenting humans, are presented in the form of ROI, yet they are expected to outperform humans — which feels like a double standard. This places disproportionate expectations on robotic experiences, thereby creating a hurdle rate that is difficult for a startup to overcome. Question for customers: Were personal computers and productivity software evaluated on an ROI basis? Many enterprise customers purchased computers and gradually loaded them up with software, leading to greater value-add over time. Greater utilization of computers led to greater productivity, but it didn’t happen overnight. Why are robots being held to a higher standard?
STARTUP: How does your robot fit into the big picture? Many roboticists and entrepreneurs are so fixated on the sexiness of their robots that they lost sight of the broader ecosystem in which they will operate. Many spend time working in the hotels, fulfillment centers, restaurants, and hospitals they hope to automate, but pay too much attention to the task rather than the other people that make up the operation. If the goal is to make the operation more streamlined and efficient, then the value proposition should be described by contrasting the economics of the operation with, and without the robot helpers. If the economics as a whole are not significantly impacted for the better, why even bother with the robots in the first place?
STARTUP: Did I forget to say that robots are SLOW? Human labor to the rescue! Navigating unstructured environments continues to be a major technical challenge for robots. Tasks that have not been automated to date are typically in such complex environments where humans thrive. Examples include elderly and child care, classrooms, kitchens, low-volume manufacturing, and order fulfillment. Technologists need to be creative around not only making the robots faster, but finding creative ways to simplify the problem so that the robots aren’t competing head-to-head with human workers, but rather making humans faster and their jobs easier. For example, Locus Robotics is automating order fulfillment by cooperating with humans; while making their jobs easier
INVESTORS: Get engaged early with the startups. Many robotics startups come as a result of controls engineers, deep learning, and computer vision experts coming together. Investors can play a key role with the startups by tee’ing up customers and helping the startups iterate on product based on customer interactions. Unlike SaaS, mobile, or social startups, where investors have the luxury of observing what “sticks,” robotics startups can benefit greatly from investors’ experience in networks as they grow their teams and their products. Being on the verge of a new industry, I would expect these investors to be handsomely rewarded.