top of page


Behavior-based robotics originated in the mid-1980’s as a reaction to the lack of progress with planning-oriented artificial intelligence. Since the early success of  Shakey (the first artificial intelligent robot) in 1969, roboticists had become consumed with deliberation and planning. The belief was that the only way to make a robot do anything intelligent, even just walk across a room, was to plan every step and reason about every object. Planning is computationally hard and time consuming. And overkill: a cockroach is pretty dumb yet it seems that the majority manage to quickly flee or hide in a room. Cockroaches, fish, frogs, sheep– they aren’t planning or performing deliberation. They are reacting, such as reflexively running away from cues of a predator (turning the lights on makes most species of cockroaches runaway) or stay near an area with food until the amount of effort begins to exceed the food that is being found (foraging). These reaction packages are called behaviors by animal ethologists and are combinations of sensing and acting. For example when the Gunslinger in the original Westworld comes after Richard Benjamin, it uses Benjamin’s heat and audio signature to track him. Yul Brenner’s Gunslinger isn’t  really smarter than a mosquito, just deadlier. Same goes for the Sentinels in the X-Men; they aren’t too bright but they are smart enough to be deadly. While insects and animals aren’t necessarily smart enough to take over the world Skynet-style, they are intelligent. And humans use that same behavior-based intelligence as part of the lower brain. So a new cadre of roboticists started focusing on this lower, behavior-based style. The most famous is MIT’s Rod Brooks and his work that led to the roomba.

bottom of page