In the near future, vehicles will be equipped with a variety of sensors, computers, and communication systems, for example to implement driver assistance functions. On the civilian market, this development is largely driven by a growing interest in safety and comfort. Additionally, the military has an increasing demand for situation-awareness capabilities in their vehicles. These trends have become well established in the scientific community. However, current research on these topics is often hampered by the lack of an adequate, configurable testing and demonstration platform.
A versatile research vehicle is needed that provides testing and analysis functionalities for a wide range of sensors and various operating scenarios. Depending on the investigated scenario such a vehicle can even play different roles, for example a leader or follower in a convoy, a single vehicle in traffic, or a command and control vehicle. Example applications for its sensors are obstacle detection and avoidance, traffic monitoring, acquisition of 3D data, change detection, as well as target location, target tracking, target designation and target handoff between vehicles. It is even possible to simulate a network of multiple vehicles and their interaction.
Fig. 1: The MODISSA platform enables mobile sensor data acquisition, sensor data analysis and mobile demonstrations.
MODISSA (Mobile Distributed Situation Awareness) is the IOSB's realization of an experimental platform for hardware evaluation and software development in the above contexts of automotive safety, security, and military applications. It is based on a Volkswagen van VW T5 that has been equipped with a broad range of sensors and contains hardware for complete raw data capture, real-time data analysis, and immediate data visualization on in-car displays (Figure 1). The VW van carries several sensors on a roof rack, and a power supply as well as operational electronics inside. The sensor configuration can be adapted to the needs of the respective study. The electronics, including several PCs, are located in a rack behind the driver's seat. A row of passenger seats behind it is arranged for people operating the system or watching real-time processing demonstrations. The power for the sensor system is provided by four high-capacity Li-ion batteries that are stored in a box in the back of the van. This power system has sufficient capacity for several hours of independent operation.
Control of the sensors and data acquisition are performed by PCs in a rack inside the van. Three types of PCs are included: A control PC, several recording PCs, and a powerful processing PC. For georeferencing and time synchronization, an Applanix inertial navigation system is built into the van. Its IMU (Inertial Measurement Unit) is mounted on the same mounting plate as the main sensors to minimize errors caused by the plate's elasticity. GPS antennas are positioned on the roof near the front and back of the vehicle.
Fig. 2: Mobile 3D model acquisition with the MODISSA platform. The data acquired with two Velodyne laser scanners (blue and orange for visualization) constitute highly accurate 3D representations of the vehicle's vicinity. 3D data can be accumulated for wide area models and can be fused with imagery acquired with the camera sensors of the MODISSA platform. See also: test data "MLS 1"
The sensors are mounted on plates fastened to two cross bars of a standard roof rack. The current sensor configuration includes two rotating Velodyne laser scanners, an omnidirectional camera, and two cameras on a pan-tilt unit (one visual-range camera and one microbolometer infrared camera). The laser scanners are located ahead of both roof rack bars over the front corners of the vehicle roof, and are positioned on a wedge with a 25 degree angle to the horizontal, sloping to the front outside at a 45 degree angle. This configuration guarantees a good coverage of the roadway in front of the car and allows scanning of building facades alongside and behind it (Figure 2). A vertical plate between the laser scanners serves to shield these from mutual direct laser radiation. The omnidirectional imaging system is directly fastened to the roof rack bar near the back of the vehicle, and the pan-tilt unit is located in the center of the vehicle between the rack bars to reduce interference with the two laser scanners.
Fig. 3: Texture mapping and visualization of a point cloud of an urban area.
An initial version of the MODISSA platform has been realized. Currently implemented applications include data acquisition and interaction paradigms with the sensors, such as looking around through the roof-mounted omnidirectional sensor with a head-mounted display when sitting inside the car. We are currently realizing more encompassing applications on the MODISSA platform, such as pedestrian detection, target detection and target handoff applications, as well as wide-area 3D model acquisition. An example is shown in Figure 3: 3D point clouds - like those acquired with MODISSA - can be supplemented with RGB- or IR-image data (captured with MODISSA itself or with other platforms).
Borgmann, B., Schatz, V., Kieritz, H., Scherer-Klöckling, C., Hebel, M., Arens, M., 2018. Data processing and recording using a versatile multi-sensor vehicle. ISPRS TC I Mid-term Symposium "Innovative Sensing - From Sensors to Methods and Applications". ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume IV-1, pp. 21-28. [doi: 10.5194/isprs-annals-IV-1-21-2018]