Ken Goldberg

Beyond the Uncanny Valley: Our Fear and Fascination with Robots

Engineers, animators, and designers apply the concept of the Uncanny Valley to technologies from AI to Robots to Siri. In 1919, a year before the word “robot” was coined, Sigmund Freud published an influential essay tracing the concept of the Uncanny back to the Renaissance. Goldberg illustrates this history with art that explores the shifting borders between the digital and the natural, including his Emmy-nominated short doc film that explores our collective fear and fascination with robots, the most human of our machines.

Brainstorming At a Global Scale

To brainstorm at the scale of social media, we can use techniques from an unlikely source: Robotics. Goldberg presents recent results on social innovation and collective brainstorming work with the U.S. State Department, General Motors, and the State of California.

Putting the "Turing" into ManufacTuring: Recent Developments in Algorithmic Automation

Automation for manufacturing today is where computer technology was in the early 1960's, a patchwork of ad-hoc solutions lacking a rigorous scientific methodology. CAD provides detailed models of part geometry. What's missing is formal models of part behavior and frameworks for the systematic design of automated systems that can feed, assemble, and inspect parts. "Algorithmic Automation" introduces abstractions that allow the functionality of automation to be designed independent of the underlying implementation and can provide the foundation for formal specification and analysis, algorithmic design, and consistency checking.

Too Close for Comfort?: AI, 5G, IoT, Robots, and Privacy

Prof. Goldberg will present recent advances in artificial intelligence, robotics, and networking and the potential dangers they raise in terms of security and privacy. Goldberg will illustrate these issues in the context of advances in audio, video/face recognition, and data surveillance, including from his own research on robot learning, Freud's concept of the Uncanny, and art projects including his Whitney Museum art installation and the concept of "Respectful Cameras".

Shortly after the 1918 pandemic, the word "robot" was coined in a play about mechanical workers organizing a rebellion to defeat their human overlords. A century later, amid rising economic inequality and xenophobia, we are immersed in a new global pandemic. In parallel, emerging advances in Artificial Intelligence and Robotics, fueled by governments, corporations, and venture capital, disrupt labor, trade, and political stability. As claims are made about “superintelligence” and existential threats to humanity, new questions arise about the distinctions between humans and machines.

Asserting that "humanity still has a few good years left," Goldberg draws on art and literature from Ovid's Pygmalian to the Golem, through Von Kempelen's "Mechanical Turk" (1770), E.T.A. Hoffman's Sandman (1816), Mary Shelley's Frankenstein (1818), Sigmund Freud's Uncanny (1919) and Masahiro Mori's Uncanny Valley (1970), to contextualize our contemporary fear and fascination with AI. Goldberg will also describe his own artwork that explores the boundary between the natural and the artificial, such as the Telegarden (1995-2004), a living garden tended by 100,000 visitors operating a robot over the Internet, and his new project, AlphaGarden (2020-), that challenges AI to a game with nature.

Humans have grown food for over 10,000 years. As the climate changes and the global population seeks fresh and healthy nutrition, advances in robots are being applied in agriculture. This talk will review recent progress including John Deere's use of drones to fine-tune fertilizer delivery and EarthSense mobile robots that roll beneath leaf canopies to closely monitor plant properties that optimize breeding. I'll share results from RAPID, a USDA-sponsored project developing Robot Assisted Precision Irrigation Delivery, and two projects that incorporate art and research: TeleGarden (1995) where over 100,000 people remotely collaborated to tend a living garden, and AlphaGarden (2020), where simulation and measurements from a living garden are being combined to train a robot to sustain a diverse polyculture garden. This talk will explore how the the Agricultural and Industrial Revolutions can be extended by the AI Revolution.

Despite 50 years of research, robots remain remarkably clumsy, limiting their applications in home decluttering, warehouse order fulfillment, and robot-assisted surgery. The First Wave of grasping research, still dominant, uses analytic methods based on screw theory and assumes exact knowledge of pose, shape, and contact mechanics. The Second Wave is empirical: purely data driven approaches which learn grasp strategies from many examples using techniques such as imitation and reinforcement learning with hyperparametric function approximation (Deep Learning). The New Wave is based on hybrid methods that combine analytic models to bootstrap Deep Learning models, where data and code is exchanged via the Cloud using emerging advances in cloud computing and big data. I'll present our lab's work on the Dexterity Network (Dex-Net), an emerging New Wave approach that allows robots to grasp a broad range of novel objects.

To improve patient care, a new generation of robots is being developed to actively assist surgeons in the operating room. These robots can autonomously perform tedious subtasks such as suturing and debridement to improve consistency and reduce fatigue, analogous to advances in automotive anti-skid braking and intelligent lane-keeping systems. New companies and labs are using Recent advances in AI and deep learning are being applied by researchers and companies worldwide to develop this new generation robots, which also opens the door to long-distance tele-surgery.

In this talk I'll present recent advances from our lab including novel hardware and software with applications to cutting, suturing, palpation, dissection, retraction, debridement and a recent result -- "Superhuman Peg Transfer", where a robot autonomously performs a standard surgical task with accuracy and speed on par with a surgeon and with significantly more consistency.