Berkeley Robotics highlighted at TechCrunch event

by Saemmool Lee

Researchers and entrepreneurs in the fields of artificial intelligence (AI) and robotics emphasized the social impact of the technologies at the day-long “TechCrunch Sessions: Robotics + AI 2019” event on April 18 in Zellerbach Hall, co-sponsored by CITRIS and the Banatao Institute.

Ken Goldberg, director of CITRIS People and Robotics Initiative (CPAR) introduced the term “inclusive intelligence” that Berkeley researchers have been discussing. “It’s the idea that a lot of AI has been giving certain people a lot of advantages and a lot of people disadvantages,” says Goldberg. “What we want to do is start to develop a new approach to artificial intelligence that is inclusive in three ways.”

The first is inclusive of disadvantaged populations who have been underrepresented or left out, says Goldberg. The second is inclusive of many disciplines, because the issues about intelligence run across fields. The third is inclusive of both human and machine intelligence, which Goldberg says are complimentary. He refuted the concept of “singularity,” which is the popular notion that machine intelligence is going to supersede human intelligence. “I’m more interested in the idea that we will use AI as a tool, as a way to enhance human abilities,” he says.

Many major universities are interested in AI, but Berkeley is uniquely positioned, says Goldberg. “One thing that we have here is a sense of commitment to the public and to the public good,” he says. “Now ultimately, diversity is one of our greatest strengths. Diversity is such an important aspect, really at the core of machine learning.”

Alice Agogino, CITRIS researcher and Berkeley professor of Mechanical Engineering, introduced Squishy Robotics, a startup that she founded. The company provides rapidly deployable mobile sensing robots for disaster rescue, remote monitoring, and space exploration. Agogino demonstrated how the robots land and provide data at a disaster training facility and a fire department training center. “Machine learning will allow us to improve our analytics over time as more customers use our robots,” she says.

Squishy Robotics is a spin-off of Agogino’s research with NASA and Berkeley to provide probes to go out into outer space and complete scientific missions. Agogino was then inspired to apply the technology to earthly applications. Her team conducted many interviews with first responders, given that the lives of first responders who had lost their lives saving other people’s lives might have been saved with better situational awareness before entering a disaster zone. The interviewees wanted ground-level sensors immediately in a hazardous environment before the team arrives at the scene, says Agogino.

Randy Katz, UC Berkeley Vice Chancellor for Research and former interim CITRIS director before its founding in 2001, at the dais alongside TechCrunch Editor-in-Chief Matthew Panzarino.

“Our rapidly deployable mobile sensor robots are designed to save lives, reduce cost and risk, and increase effectiveness of emergency response,” says Agogino. “They can survive a high drop into a disaster zone, provide life-saving information to first responders. They can also work as co-robots with human partners on the ground as teams arrive at the scene.”

Trexo Robotics, a mobile robotics device company in Canada, demonstrated how its device helps children with disabilities to walk. The device is approved in the U.S. and Canada as a therapeutic tool and being used in children’s hospitals and homes. Children who were otherwise only able to take a couple of steps are taking more than a thousand steps in a single session with the device, says Manmeet Maggu, founder and CEO of Trexo Robotics. “This is just the beginning for us,” says Maggu. “We will use our experience and expertise, then build functional system for adults with disabilities as well as for the elderly.”

Meanwhile, in Zellerbach’s vast lobby, robotics startups presented how their products address the social needs of diverse groups.

Robotic animals that respond to emotional cues were also on display. Tombot, a startup that provides robotic companion animals designed for seniors with dementia, displayed its robot puppy. The product mimics the looks and behaviors of a real dog and responds to where and how they are being touched. Qoobo, a company that provides therapeutic robots in the form of a cushion with a tail, also demonstrated its product. The cushion waves when caressed and its tail swings when rubbed in the way that animals do, so that it can provide a sense of comfort for customers when they are stressed.

Back on stage, CITRIS researchers stressed the need to think critically about the AI and its roles.

“I don’t like the term AI because I don’t think we know what intelligence is yet,” says Michael Jordan, CITRIS researcher and Berkeley professor of Electrical Engineering and Computer Sciences (EECS). “What’s changed in the last 40 or 50 years is that the methods have gotten a little bit better,” he says. “But we haven’t come back to really revisit what the problem is.” Finding the ability to imitate a human, which hasn’t really been examined, is not the right perspective in thinking about intelligence, says Jordan.

Jordan offered a hypothetical example: Suppose you are a Martian computer scientist looking down at the planet Earth to learn about intelligence. You might see humans and minds, and say, “That’s obviously intelligent and we should mimic that.” However, humans are pretty specific to Earth and this intelligence may not be relevant on Mars. Moreover, it’s hard to understand brains and minds – even a single neuron is extremely complex, and we’re not even close to understanding in what it is really doing. “You’ve got to think out of the box a little bit,” says Jordan.

CITRIS researcher Anca Dragan, assistant professor of Electrical Engineering and Computer Science at UC Berkeley.

Anca Dragan, CITRIS Researcher and Berkeley professor of EECS, says that she worries about “over trust,” whereas many people tend to worry about how under trusting the technology might lead to lack of adoption. “I worry about the fact that we tend to anthropomorphize these agents, which means that if they do a little bit, we tend to think that they can actually do a lot,” says Dragan. “I don’t want to be relying on the robot to be doing things that are outside of its capability.”

“We have a long way to go in the integration of artificial intelligence and robotics,” says Randy Katz, UC Berkeley Vice Chancellor for Research at a welcome and introduction. “We think Berkeley as one of the epicenters of the artificial intelligence and robotics world, and we are really glad to have the world’s experts on this stage to engage with you today on the cutting edge of that interface between artificial intelligence and robotics.”

It was the second time that the TechCrunch’s Session event was held in the Zellerbach Hall at Berkeley. “Let’s make this a permanent fixture of the calendar,” says Katz.


Featured Image: CITRIS People and Robots Director Ken Goldberg in conversation with CITRIS researcher Michael Jordan.
Images: TechCrunch