NILS JOHN NILSSON (1933-2019)

BY RICHARD FIKES
SUBMITTED BY THE NAE HOME SECRETARY

NILS JOHN NILSSON, an American computer scientist and an early researcher in artificial intelligence (AI), died April 23, 2019, at his home in Medford, Oregon, after a long illness. He was 86.

From 1961 to 1985, Nils conducted research at the Stanford Research Institute (SRI), where he led the development of the world’s first mobile intelligent robot. In 1985, he joined the Stanford University faculty as chair of the Computer Science Department, a role he held until 1990. From 1991 until his retirement in 1995, he was Kumagai Professor of Engineering in computer science at Stanford. His research there focused on building intelligent robots — machines that could perceive their surroundings, learn, plan, act, and communicate with both humans and other robots.

Nils studied at Stanford as an undergraduate and earned a doctorate in electrical engineering there in 1958. His dissertation, in information theory, explored detecting and jamming radar. After completing his Ph.D., he joined the U.S. Air Force and spent three years stationed at the Rome Air Development Center in Rome, New York, before joining SRI.

An early focus of Nilsson’s work involved neural networks, a technology pioneered by Frank Rosenblatt at Cornell University. In 1965, Nils published Learning Machines: Foundations of Trainable Pattern-Classifying Systems, one of the earliest books on neural networks. Though neural networks fell out of favor in the 1970s, they re-emerged decades later as computing costs dropped, and large data sets became more accessible. With these advancements, neural networks began rivaling multiple forms of human intelligence.

Starting in 1966, Nilsson, along with Charles Rosen and Bertram Raphael, led a research team at SRI in a pioneering effort to construct a mobile robot. They named the robot Shakey. It used an onboard sonar rangefinder and video camera to construct a model of its environment, reasoned about that environment to create an action plan, and performed a plan by sending commands to its motors and monitoring the results. That project was enormously influential in AI and was honored as an Institute of Electrical and Electronics Engineers (IEEE) Milestone in 2017. The milestone plaque states that Shakey “could perceive its surroundings, infer implicit facts from explicit ones, create plans, recover from errors in plan execution, and communicate using ordinary English,” and that “Shakey’s software architecture, computer vision, and methods for navigation and planning proved seminal in robotics” and in multiple other areas of computer science and engineering.

Although the idea of using logical reasoning to decide actions was first proposed by John McCarthy (NAE 1987), Nilsson’s team was the first to implement it in an autonomous agent. Along the way, they invented and proved the optimality of the A* search algorithm and founded the field of automated planning for robot agents. They also developed the STRIPS planner, whose action representation remains the basis for many modern planning algorithms.

Nilsson’s appointment as chair of Stanford’s Department of Computer Science in 1985 was notable because he was an outsider to the university at the time. However, he was already recognized as a leading lecturer and intellectual force in AI. That same year, he oversaw the department’s transition from the School of Humanities and Sciences to the School of Engineering.

Throughout his career, Nilsson not only taught and conducted research but also chronicled the progress of AI as a field. In the 1980s, he authored or coauthored two widely used textbooks: Principles of Artificial Intelligence (1982) and Logical Foundations of Artificial Intelligence (1987). In 2010, he authored The Quest for Artificial Intelligence: A History of Ideas and Achievements, tracing AI’s evolution from ancient philosophical concepts to modern breakthroughs.

Nilsson served as the fourth president of the Association for the Advancement of Artificial Intelligence (AAAI) from 1982 to 1983 and was a founding fellow of the organization. He was also a fellow of the American Association for the Advancement of Science and a foreign member of the Royal Swedish Academy of Engineering Sciences. His honors included the IEEE Neural Networks Pioneer Award, the Award for Research Excellence from the International Joint Conference on Artificial Intelligence (IJCAI), and the AAAI Distinguished Service Award for lifetime achievement. In 2011, he was inducted into the IEEE Intelligent Systems Hall of Fame for his “significant contributions to the field of AI and intelligent systems.”

Upon becoming the first Kumagai Professor of Engineering, Nilsson expressed his perspective on engineering:

“My attitude toward engineering is that it is not sufficient for our constructs merely to work well — they must be elegant, they must be as simple as possible, and they must be well organized. In short, they must be well composed! There is a very practical reason for adhering to these aesthetic standards. The reason is that if our research is ever to be used by others it must also be able to be taught. People do not easily remember nor learn overly complicated things, and if no one remembers or learns about the results of our research, our efforts will have been wasted. But people are especially attracted to elegant engineering ideas just as they are to elegant art. So, in my efforts to invent more versatile and useful robots, I always try to seek the most elegant designs.”

Beyond his technical achievements, Nilsson was equally beloved as a mentor and colleague to many, including myself. I worked closely with Nils at SRI on the Shakey robot project, and later in my career, I knew him as a fellow professor in Stanford’s Computer Science Department. When I arrived at SRI as a new Ph.D., Nils quickly became a model scientist, colleague, innovator, life coach, and gentleman to me. He remained a prominent and highly influential presence in my life, as well as the lives of many others, throughout his career.

Karen Myers, a former student of Nilsson, wrote:

“Nils deservedly was awarded every major accolade in the field, serving as a testament to his outstanding technical leadership and impact. I think what is even more special about him is how deeply he affected people on a personal level. His enthusiasm was infectious and provided a level of optimism and energy that propagated readily to those nearby. His students and colleagues at both Stanford and SRI adored him. Almost without fail, the mention of his name brings a smile to people’s faces and a fond story about some interesting, shared experience or exchange with him. He is deeply missed. I’ll be forever grateful for the many wonderful memories and the inspirational mentorship that he provided to me and so many other people in the AI community.”

Nils is survived by two children from his first marriage to Karen (Braucht) Nilsson, who died in 1991; his wife of 27 years, Grace Abbott; by her four sons from a previous marriage; and twelve grandchildren. His two children are daughter Kristen Nilsson Farley of Los Angeles and son Lars Nilsson of Piedmont, California.