-
Notifications
You must be signed in to change notification settings - Fork 142
Technical Evaluation
In this section, we discuss, amongst others, the bandwith, data throughput, the computation time and possible delays. Therefore, we show some calculation examples and accompany them with experimental evaluation data.
Let us assume we want receive video data from a high resolution camera with 720p (1280 x 720). Thus, a single image from the video stream has 921,600 px. We further assume the image will be decoded with the RGB8 standard from ROS, which means each pixel is decoded as three 8 Bits values, one for each base color. Thus, in total a single image has 22,118,400 Bits. With transmission rate of 30 Hz, we would require to transport 663,552,000 Bits/s, thus approximately 663 MBit/s. Such high data transmission rates are quite unlikely to reach, for example the Raspberry Pi reaches with the built-in WIFI approximately 36 Mbit/s. With a WIFI adapter we might reach transmission rates of approximately 80 MBits/s (Source). However, this is still not enough to transport HD images directly over WLAN if transmission rates of 30 Hz want to be reached. Thus, the images have to be compressed, for example using JPEG which has a compression ratio of approximately 10:1. This leads then to a required data transmission rates of 66 MBit/s. By reducing the transmission rate to 5 Hz, we can even ensure to get all image data via the WLAN connection of the built-in WIFI. Of course we are aware such as H.264, H.265 are much better suited for bandwith reduction. However, currently only JPEG and PNG compression is supported by the ROS image standard.
The speed of the data processing for the incoming data is highly dependend of the processing power of your mobile device. One of the most computational complex operations is the transformation of the occupancy grid message into a Bitmap for displaying the map in the Vit Tab. Depending on the size of the Occupancy Grid, this procedure might take a while.