Adaptive ajax-based streaming video for the irobot create platform for use in buildings with infrastructure mode 802.11 networks

摘要

An adaptive, Internet streaming video service for use in consumer grade Web-enabled telerobotic systems is presented. The system, based on the FFMPEG multimedia framework, is designed to stream video at different dynamically controlled frame rates and video quantizer scales as a function of current wireless data rate. Video is captured from an onboard camera connected to an embedded single board computer (SBC) through an IEEE 1394 (Firewire) serial interface, all set within the robot's payload bay. As a mobile robot engages in indoor tasks that span long distances in corridors on multiple floors within buildings with integrated 802.11 infrastructure mode wireless service, different signal strengths may be observed by the robot's wireless transceiver due to free space path loss and attenuation from building infrastructure, which effectively varies the WLAN data rate. In the 2.4 GHz band used by 802.11b and g, path loss from drywall can be 4 dB, brick wall 8 dB, concrete wall 10–15 dB, and 20 dB through stucco wall that uses diamond lath. An 802.11 transceiver will step down its data rate as signal strength decreases. As a result, the rate and quality of real-time transmission of video frames from a mobile robot must adapt to a changing signal strength to prevent jitter and packet loss which is undesirable in mobile telerobotic systems, as it is essential for the human controller to have an accurate picture of the robot's position and orientation at all times to avoid collision. In this paper, we present an adaptive system that varies the frame transmission rate or the quantizer scale in real-time, as a function of the robot's current WLAN transceiver data rate. The numerical method employed is based on acquiring the effective data rate required to stream video frames from 3 to 30 fps at different quantizer scales, fitting this data to a polynomial model, and using an iterative Newton-Raphson solver to solve the inverse problem to perform rate adaptation. Performance results based on variable frame rate at a fixed quantizer scale and variable quantizer scale at a fixed frame rate are presented.