Synchronization of multiple rates Concept. Often multiple sensors are being used to sense the environment for mobile robotics. While an IMU sensor provides data samples at a very high rate (e.g., 500 Hz), laser scans are availabe at a much slower frequency (e.g. 10Hz) determined by the revolution time. Then the challenge is, how to deterministically fuse sensor data with different frequencies. This problem is depicted in Figure 5. Figure 5: How to deterministically process multi-frequent sensor data. Due to scheduling effects, the callback for evaluating the laser scan might be called just before or just after an IMU data is received. One way to tackle this is to write additional synchronization code inside the application. Obviously, this is a cumbersome and not-portable solution. An Alternative would be to evalute the IMU sample and the laser scan by synchronizing their fre- quency. For example by processing always 50 IMU samples with one laser scan. This approach is shown in Figure 6. A pre-processing callback aggregates the IMU samples and sends an aggregated message with 50 samples at 10Hz rate. Now both messages have the same frequency. With a trigger condition, which fires when both messages are available, the sensor fusion algorithm can expect always synchronized input data.
