We’re edging ever closer to the moment when robots toiling in industrial settings break free from their metal cages and begin working alongside their human colleagues as efficient and helpful companions.
While nobody expects robots to begin organizing for shorter working hours or to join their human colleagues for Happy Hour, a growing number of robotics experts are envisioning factory floors with robotic assistants performing tasks that human workers find either hazardous or excessively burdensome, particularly in fields such as automotive, aircraft and heavy equipment manufacturing.
A delicate new world
Life is grim for today’s industrial robots, which are generally relegated to performing repetitive tasks and manipulating heavy tools and materials while humans focus on work requiring delicacy, detail and creativity. But as robots become capable of performing finer and more precise tasks, it appears likely that they will soon find roles on the main factory floor, not to replace humans but to relieve their co-workers of mundane, repetitive and time-consuming tasks.
Dr. Julie Shah, an assistant professor of aeronautics and astronautics at the Massachusetts Institute of Technology (MIT), observes that robotic workers could someday play a major role in global manufacturing, helping companies that produce cars, airplanes and other complex products more efficient by transferring dull and unimaginative tasks to machines.
“If the robot can provide tools and materials so the person doesn’t have to walk over to pick up parts and walk back to the plane, you can significantly reduce the idle time of the person,” says Dr. Shah, who leads the Interactive Robotics Group in MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL)
An important key to making robots helpful partners is software that will allow the machines to quickly and easily adapt to different conditions and situations. Since a co-worker robot is likely to work with numerous shift workers as well as sick time and vacation fill-ins, it needs to be able to seamlessly adapt to each human partner’s personal style and preferences.
To meet this need, Dr. Shah and her co-researchers have created an algorithm that’s designed to allow robots to quickly learn and adapt to a human worker’s task preferences and routines. The researchers are currently testing the algorithm in simulations that place robots and humans within a single working environment.
“It’s an interesting machine-learning human-factors problem Dr. Shah says. “Using this algorithm, we can significantly improve the robot’s understanding of what the person’s next likely actions are.”
Dr. Shah envisions a real-life manufacturing setting in which robots and humans participate in an initial training session away from the factory floor and then continue refining their interactions on the job. Once the robot learns a particular worker’s work habits, the machine can be programmed to recognize?and adapt to the preferences of?that same person whenever it encounters the individual on the factory floor. Dr. Shah notes that many production workers already wear radio-frequency identification (RFID) tags?a technology that could be used to help robots quickly discover and identify specific human workers.
To test their algorithm and overall concept, Dr. Shah and her co-workers turned to a robotic arm designed by Swiss company ABB. Her team named the robot ABBIE. Dr. Shah says the robot can be easily instructed to perform specific assistive tasks, such as fastening bolts or applying sealant to holes, depending on a human’s preferences.
To allow ABBIE to predict a human worker’s actions, the researchers created a decision tree-oriented computational model. Each branch along the tree represented a specific choice a mechanic or technician might make, such as deciding to hammer in a bolt after applying a ring of sealant, or applying sealant to the next hole and postponing the hammering operation until later.
To be a supportive co-worker, ABBIE needed to be able to anticipate such decisions.
“If the robot places the bolt, how sure is it that the person will then hammer the bolt, or just wait for the robot to place the next bolt?” Dr. Shah asks. “There are many branches.”
Once the model was completed, the researchers performed a series of experiments, training the robot to observe a human worker’s chain of preferences. Once the robot learned a person’s preferred order of tasks, it quickly adapted to the situation, Dr. Shah says, either applying sealant or fastening a bolt in accordance with the individual’s specific preference.
Colleague is a Hiro
MIT isn’t the only organization hoping to bring robotic co-workers to factory floors. Tecnalia, a Spanish technology research center, is planning to transform a Japanese research robot into a steadfast industrial working companion.
The robot, named Hiro, is currently used in Japan for various robotics research activities, including working alongside humans. Developed by Kawada Industries, Hiro has a humanoid appearance in its upper part and a robot’s body below. Kawada, like many other Japanese firms, believes that equipping robots with a human appearance encourages their entry into the workplace and their interaction with the workers.
Hiro’s upper body features a head, a trunk and two extremities that terminate into a hand. The system also includes four cameras that function as eyes: two on the face and one on each hand. Hiro’s lower half features a wheeled mechanism that enables the machine to move around a factory floor and between workstations.
Tecnalia is currently upgrading Hiro’s intelligence in an effort to allow the robot to perform different tasks covering a range of industrial scenarios and automatically adapt itself to the work needs of specific manufacturers. Tecnalia is currently working with European aircraft manufacturer Airbus to evaluate Hiro’s potential for assisting human aircraft builders. Representatives of both organizations recently traveled to Japan where they were able to jointly study Hiro and its ability to perform factory floor tasks.
What’s special about Hiro is that it is a social robot that’s built to share working space with people in conditions of absolute safety, says a Tecnalia spokesperson.
“In the event the robot comes into physical contact with a human, it is programmed to stop automatically,” the spokesperson notes.
Peering into the future, Dr. Shah envisions human-robot workteam applications that extend far beyond the factory floor. She notes, for instance, that robotic assistants may be programmed to help physicians, nurses and other caregivers in a variety of medical settings. A robot could be trained, for example, to work inside an operating room, anticipating a surgeon’s needs by handing over the appropriate scalpels, clamps, gauze and other tools the moment they are needed. Inside a long-term care facility, such a robot helper could deliver meals and clean up spills and other messes, freeing staff members for tasks that require direct interaction with patients.
Human-robot work teams are likely to arrive gradually over the next several years as system costs fall, computer and robotics technology advances and an array of sophisticated algorithms become available to address the needs of different industries and work tasks.
“We have hardware, sensing and can do manipulation and vision,” Dr. Shah says. “But unless the robot really develops an almost seamless understanding of how it can help the person, the person’s just going to get frustrated and say, ‘Never mind, I’ll just go pick up the piece myself.'”