kidnapped robot problem
The situation where an autonomous robot is carried somewhere and as a result has trouble determining its own location.
kidnapped robot problem: lost after being moved without warning
A kidnapped robot is one that has been physically relocated without updating its internal position estimate. The robot suddenly finds itself in a location it did not navigate to under its own control. Since most autonomous robots build their position knowledge through odometry, SLAM (simultaneous localization and mapping), or GPS, being picked up and placed elsewhere creates a mismatch: the robot's computed location is wrong, but it has no sensor input to tell it that movement occurred.
The core issue arises because wheeled and tracked robots typically estimate position by integrating encoder counts from their drive motors. If a robot travels 2 meters forward, it records that 2-meter displacement. But if a human carries the robot 5 meters to the side, the encoders register zero motion. The robot's internal map still shows it at the old location, creating a localization error that can cause navigation failures, collision risks, and task incompletion.
Mobile robots encounter this problem most often in warehouse and factory environments where human staff move equipment without coordination with the robot's control system. A robot being loaded onto a truck or repositioned by maintenance workers will lose its sense of where it is. Even brief manual corrections, like an operator nudging a stuck robot sideways, introduce localization error.
Recovery and Prevention
Robots equipped with global positioning systems or overhead localization infrastructure (fiducial markers, UWB networks, lidar reflectors) can detect and correct kidnapping more readily. A SLAM-based robot may relocalize by observing distinctive features in its environment, though this can take time and may fail in featureless spaces. Some robots include manual reset buttons or docking stations that announce the robot's precise location to its navigation system.
The term "kidnapped robot problem" originated in probabilistic robotics research during the 1990s and remains standard in localization literature. It highlights a fundamental constraint: autonomous navigation systems must assume their sensors are their primary source of truth about movement. When that assumption breaks, recovery depends on absolute localization methods rather than dead reckoning alone.