SLAM, Simultaneous Localization and Mapping, is the computational problem at the heart of autonomous robots. It asks how a robot dropped into an unfamiliar place can figure out both where it is and what the place looks like, when it has neither a map nor a known starting position to anchor it. The influential 2006 tutorial by Hugh Durrant-Whyte and Tim Bailey, “Simultaneous Localization and Mapping: Part I,” published in IEEE Robotics and Automation Magazine, states the question precisely: whether “it is possible for a mobile robot to be placed at an unknown location in an unknown environment and for the robot to incrementally build a consistent map of this environment while simultaneously determining its location within this map.”
The difficulty is that the two halves of the problem depend on each other. To build a map you need to know where you are, so you can place what you see in the right spot. To know where you are you need a map to compare your sensor readings against. SLAM is the software answer to this chicken-and-egg coupling: it estimates the robot’s trajectory and the map together, treating both as unknowns and refining them as new sensor data arrives.
In practice a SLAM system fuses noisy measurements over time. Wheel odometry, inertial sensors, lidar scans, and camera images each carry error, and the algorithm must combine them into a single consistent estimate, correcting accumulated drift when the robot revisits a place it has seen before, an event called loop closure. The early formulations leaned on probabilistic filters; later work moved toward graph-based optimization, but the framing from the Durrant-Whyte and Bailey survey, joint estimation of location and map, remained the conceptual core.
SLAM is what turned mobile robots from machines that follow pre-surveyed routes into machines that explore. It underpins the navigation stacks shipped with robotics frameworks like ROS, the autonomy in vacuum robots and warehouse vehicles, and the spatial tracking inside augmented-reality headsets. Wherever a device must know where it is in a world nobody mapped for it in advance, some form of SLAM is doing the work.