Amid the swarms of students engaged in higher learning on the University of California, Berkeley campus, student Ajay Tanwani’s attention was fixed on the pint-sized figure in the middle of his basement laboratory trying to figure out the difference between a screwdriver and a roll of electrical tape.
The figure — a plastic robot about waist high — eventually reached down and plucked a screwdriver from a clutter of items and deposited it in one of a row of boxes, each marked with a code.
"Yeah, you just saw the future right here," Tanwani said, "That’s what we are going for."
This future — a robot sorting items scattered on the floor at a snail’s pace — might not appear quite as advanced as vintage science fiction forecasted for 2018, yet it was doing something most robots don’t do — it was learning and reasoning. When the robot became overwhelmed by the cluster of items before it, it reached out to the humans in the room via computer software to ask for advice.
"When they are stuck in a tricky situation, they can then request a supervisor on the remote side to provide some kind of assistance in how to overcome this problem," Tanwani explained.
Tanwani banged out an instruction on the computer and the robot went back to work, picking out a piece of foam from the floor amid other detritus and dropping it in its respective box. Tanwani said he and fellow engineering students are incorporating cloud-based computing into operating the robots, a concept known as fog robotics, in which the devices can seek instructions from an information ether, which can also be networked with other computers. Science fiction never saw that coming.
Local
"So the idea being the robot will operate using its own computer on the edge," said Cal robotics professor Ken Goldberg, "But every once in a while it needs to go up into the cloud to download some new information."
The robots under development at Cal aren’t the kind that terrorize planets or crush enemies with lasers, but rather they take on practical tasks they may eventually assume from their human counterparts — things like cleaning the garage or sorting packages in a warehouse.
"We look at applications ranging from warehouse automation, which is a big focus right now, to home robots and surgical robots," Goldberg said.
Of course with each robotic technological victory, comes a new, and well-founded fear of robots putting the humans out of work. Students in Cal’s program recently presented a warehouse sorting machine to the honchos at Amazon. In what resembled a pair of arms coming out of a wall, the robot was able to dig into a box of various items and sort everything into separate receptacles based on appearance.
Yet Goldberg said any jobs that require a fine touch — from plumbers to mechanics to health care workers — are probably robot-proof, at least for now.
"I think it’s important," Goldberg said, "Also a responsibility to reassure people that humans have many good years left."
In a nearby building, engineering student Andrea Bajcsy was maneuvering a single multi-jointed robotic arm clutching a coffee cup. By enabling the robot’s software learning function, Bajcsy’s was able to influence the motion of the robot’s arm simply by gently nudging it.
"The robot can learn your preferences based on interacting with you in that context and perform the task you want it to do even better," Bajcsy said.
Bajcsy is focused on developing robots to work alongside humans, taking cues from their human counterparts and incorporating those skills into their learning. She sees a vast potential for robotics to help out in home environments.
"Maybe you’re elderly or you’re disabled and this robot is helping you with your day-to-day functions," Bajcsy said.
The yet unrealized potential of robots has inspired both hope and fear for the future. Some, like Tesla founder Elon Musk, have raised warnings that unregulated artificial intelligence and robotics threaten not only jobs but humankind itself. Among the dystopian scenarios floated: a network of autonomous robots could turn on their creators and wage war against humans.
Cal engineering professor Anca Dragan, who studies robotics ethics, says such a scenario is mostly far-fetched.
"There could be malicious actors that specify the wrong things," Dragan said. "But for the most parts the only way that something will revolt against us if someone gave it a reward function that incentivized that."
Goldberg said despite the seemingly-constant stream of innovations, the endless YouTube videos of robots doing stunning feats such as backflips, robots are still a ways off from becoming the autonomous self-starters some have imagined.
"Saying they will do everything by themselves, I think that’s still in the realm of science fiction," Goldberg said.
Back in Tanwani’s lab, the limitations of man and machine had temporarily hit the skids. The robot, suddenly perplexed by the task at hand sat motionless, its quasi-face going blank as the students nearby fiddled with the computer. Soon, it was back on track, methodically picking up rolls of tape and screwdrivers and dropping them in boxes.
The future had arrived. And it was taking its own sweet time.
