Cell-based biosensors allow to simply and selectively sense diverse chemical signals; yet their applications are limited by the minutes-to-hours timescale of gene transcription and translation. To generate a real-time output, we exploit the much faster changes in protein interaction and bacterial movement. Based on the E. coli Tar chemotaxis receptor, we developed two sensing systems: detecting DNA binding of a transcription factor via split luciferase complementation, and imaging the movement of bacteria at the single-cell level. The sensory domain of Tar can be modified to recognize different molecules, extending the applicability of the sensor. To show the advances brought by our system we built AROMA, an autonomous robot that is directly driven by the onboard biosensor. The robot detects the concentration of volatile compounds in air by imaging the bacterial response with a microscope built in-house. This enables our device to locate the source of pollutants or chemical hazards.