The discussion revolves around the challenges of building robots that can adapt to changing environments, rather than robots that rely solely on prediction and mapping.
The concept of the Uncanny Valley was created by roboticist Masahiro Mori. The graph looks at how people's affinity for robots goes up as they become more human-like, but then goes down to feeling creeped out or repulsed before potentially going back up again.
This podcast explores the different examples of things that people call robots and their level of agency, such as the ability to interact with the environment, converse with other agents, take responsibility, make decisions, and make plans. Real-world examples of robots include Roomba vacuum cleaners and self-driving cars.
Pieter Abbeel discusses the challenges of combining perception and control in robotics and how he believes treating them together produces better results than trying to separate them.
Boston Dynamics’ robot moving at five miles an hour isn't as outlandish as some critics suggest, as cars on the road moving at 75-85 miles an hour are processing data, doing machine learning and figuring out the car's trajectory in real-time, making the robot’s task easier. Special-purpose automation is a cheaper and more elegant solution to solve customer's problems, given the leveraging of core IP at Boston Dynamics can provide a great set of applications to existing product lines and customers they serve for businesses in special purpose automation, which Hyundai is now part of, following its acquisition of Boston Dynamics from SoftBank.
The emergence of robots that can mimic human behavior and characteristics raises the question: when does a robot become human? Physicist Michio Kaku argues that if a robot acts and looks like a human, it should be considered a human.
The speaker discusses the impact of robots in the industry, exploring how they can improve efficiency and require more skilled labor. However, there are concerns about automation taking away jobs from human workers leading to a dystopian future.
The ability for robots to perceive humans, including anticipating human behavior, is a crucial aspect of their development and application. Understanding how humans move and interact in various environments allows for better localization and intention recognition.
