In the next decade, many old bridges will be exposed to increasing traffic loads and destructive environmental conditions. Measurement methods like laser scanning, infrared thermography, photogrammetry, ground penetrating radar, or ultrasonic scanning are used on single robotic systems to partially support the inspectors. However, time-consuming manual inspections for crack detection, measurement, and documentation are still necessary. This paper describes the concept of an autonomous mobile robotic bridge inspection system. The proposed concept for an unmanned ground vehicle (UGV) is achieved by a trade-off of different mobile platforms, sensor systems for mapping, localization and inspection, and fist tests assessing the feasibility. We use a small concrete bridge in Freiburg (Germany) with various cracks for testing the sensors, the UGV concept, and initial tests of the mobile platform. This results in the choice of selecting the weatherproof version of the mobile robotic platform Husky from Clearpath Robotics. It is equipped with Swift Navigation's Duro real-time kinematic (RTK) system, a heading system, an inertial measurement unit (IMU), a base station, and software for semi-autonomous navigation. In the next step, we compare different sensor systems. For mapping and localization, we decide to use the 360 spherical camera Ladybug 5+ from FLIR Systems and a Velodyne VLP-16 light detection and ranging (LiDAR). High-resolution cameras allow recording damages on the bridge's surface. We perform first tests using monochrome and colour cameras. After evaluating different sensor integration concepts, we present a preliminary design of the UGV including integrated sensors.