Moravec, Hans

1948–

Robotics Innovator

A researcher in robotics at Carnegie Mellon University for the last two decades, Hans Moravec has advanced the state of the art in mobile robotics, especially with regard to providing robots with better spatial information via computer vision and other sensors. He is perhaps best known for his outspoken views on the future co-evolution of human beings and robots, and the eventual superiority of the latter.

Moravec was born in Kautzen, Austria, but lived in Canada from age five. He earned a bachelor's degree in mathematics from Acadia University in 1969, and a master's degree in computer science from the University of Western Ontario in 1971. He completed a Ph.D. in computer science at Stanford University in 1980 before moving to Carnegie Mellon, where he continues to work as the director of the Robotics Institute.

Moravec argues that what seem like the more mundane problems that must be solved before robots can interact within a real-world environment are actually among the most intractable problems of artificial intelligence (AI). Examining the evolutionary progress of biological systems, he notes that the higher functions of the human brain took very little time to develop on an evolutionary scale, while the perceptual apparatus took much longer. While we may be more impressed by a chess-playing computer, the more difficult tasks are ones that we as humans take for granted, like creating a cognitive map of the area around us or being able to quickly and safely grasp an object.

His dissertation treated the problem of moving a robot through a crowded environment, using a television picture to help guide it from one side of the room to another. This robot, called the Stanford Cart, was developed between 1973 and 1980. Although it could make it safely through the room about 75 percent of the time, it took several hours to complete the task, and it was remote-controlled by a large computer. More recent work using binocular television cameras and an approach that Moravec calls “3D occupancy grids” allows a robot to determine the layout of a real-world environment in several seconds.

Moravec is probably better known for his opinion that robots will overtake humans in the near future. In two books, Mind Children: The Future of Robot and Human Intelligence (1988) and Robot: Mere Machine to Transcendent Mind (1999), he has suggested that when robots do become superior, the most reasonable response will be for humans, at least in their current biological form, to gracefully bow out of the evolutionary process. Once these robots begin reproducing, their improvement will be exponential from generation to generation.

Based on an approximation of the relationship of neurons to computer processors, and on the exponential growth rate of processing power, Moravec estimates that we will have developed computers with processing power equivalent to humans by 2040, and that machines will far surpass human intellect in the years thereafter. This is a fairly inevitable process, he argues. Only through robots will human culture be able to thrive and spread itself across the universe.

His work spans from the near future, in which he expects robots to require procedural programming; to a middle term, when humans can co-exist with intelligent robots; to a more distant future in which biological humans are rendered extinct. He concedes that there are ways for our minds to continue on in this future, but our bodies will no longer be competitive from an evolutionary perspective. They will be replaced by robots that are able to thrive in harsh environments—possibly those that are structured like robot “bushes” with molecular-level fingers attached to increasingly larger branches.

...