The Pegasus Cam is a low-cost, low-latency network camera solution for educators, researchers, and inventors. Pegasus Cam utilizes a low-latency network stream to achieve “real-time” computer vision for typical computer vision applications such as AR-code recognition, facial detection, and person tracking. Pegasus Cam was made to be a low-cost solution to the expensive computer vision solutions available for today's robots, taking away the need to have expensive computers on the robot to perform on-board vision processing. With Pegasus Cam, computer vision algorithms can be run on a local or remote computer/server with increased processing power. The Pegasus Cam was designed to have the lowest latency possible, allowing real-time algorithms to be developed using the platform.
A paper on the development of the Pegasus Cam can be read on IEEE Xplore. This paper reviews the motivation for creating the Pegasus Cam, the methods used, and experimentation to demonstrate the low-latency capabilities of the Pegasus Cam.
The Pegasus currently comes in two forms: ROBOTIS-Mini and EveryDay. EveryDay Model COMING SOON
The Pegasus Cam utilizes a Raspberry Pi Zero W and Raspberry Pi Camera as the main hardware for its implementation. The Raspberry Pi Zero W was chosen for its integrated networking with built-in WiFi and low price. The Raspberry Pi Camera provides high quality video at 30 fps and has plug-n-play functionality with the Raspberry Pi Zero W. The specifications and pictures for the Raspberry Pi Zero W and the Raspberry Pi Camera can be seen below:
Raspberry Pi Camera ($30)
The Pegasus Cam operates on an ffmpeg streaming architecture originally developed by Silvan Melchoir, with optimizations to latency and OpenCV integration made for this version. The ffmpeg library is a highly powerful multi-media library for handling images and video streams locally and over networks. This framework allows the Pegasus Cam to seamlessly stream video over a network, and does so with low latency. The Pegasus Cam framework includes improvements to encoding, buffering, and stream management. Once the stream has been received by the web interface, it can then be integrated into OpenCV as a live camera source, allowing any OpenCV algorithm to be run just as in the case of a local camera. The full framework for the Pegasus Cam can be seen below:
This section contains videos showing the capabilities of the Pegasus Cam.
This video shows the Pegasus Cam being used to detect Aruco markers that can be used for AR applications and motion capture systems.
This video shows the Pegasus Cam being used to track faces using the Haar Cascades classifier.
This video shows the Pegasus Cam being used to track a red ball.
This video shows 2 Pegasus Cams being used for real-time video homography over a network.
For questions about sourcing, purchasing, support, setup, usage, etc. Email: email@example.com