QUASSAR at a Glance

QUASSAR (codename QSR-101) is an advanced quadruped robot aimed to provide functions in support, surveillance and  education in multiple environments, industrial, corporate, home and lot more.

It features advanced sensory systems and computational power for all of its functions. It implements a computational grid distributed in 3 nodes / boards with the following characteristics:

Cortex Node
Samsung Exynos5422 Cortex™-A15 1.8Ghz Quad-Core and Cortex™-A7 Quad-Core CPUs (Octo-Core)
* Mali-T628 MP6(OpenGL ES 3.0/2.0/1.1 and OpenCL 1.1 Full profile)
* 2Gbyte LPDDR3 RAM at 933MHz (14.9GB/s memory bandwidth) PoP stacked
* eMMC5.0 HS400 Flash Storage

Motion Node
SoC Broadcom ARM® Cortex®-A7(ARMv7) 0.9Ghz Quad-Core CPU
* Mali™-450 MP2 GPU (OpenGL ES 2.0/1.1 enabled for Raspbian)
* 1Gbyte DDR3 SDRAM

Support Node
8-bit Microcontroller Board (designed and built by me) running at 40 MHz. 4KB of internal RAM.
ADC S/H for the Front & Read proximity sensors (pseudo-Lidars)
PWM & Control for the front Ilumination Unit


The system configures a "Grid" architecture on which, the core nodes are connected through Ehternet LAN at 100Mbps, and the Support node is accessed over I2C. The main nodes are remotely accessed over WiFi independently, but both are continuously interchanging information through the LAN network.

It features two "pseudo-lidars" front and rear to perform near object detection. These sensors continuosly scan the front and rear perimeter to provide accurate information about surrounding objects.

Special software patterns and styles based on RTOS architecture have been implemented. A complete and proprietary mathematical model has also been designed to provide all of the IK functions.

It features advanced object and face recognition functions based on OpenCV, by using classifiers (Haar Cascade Classifier). It also will implement advanced AI and voice processing features based on well-known libraries like Google's TensorFlow, BlueMix, Fann etc.

The remote command and control console allows for the whole telemetry by using a gamepad, a mobile phone or my gesture capture glove (see it at: HMI Wireless Gesture Capture Gloves).

This is a project in constant evolution, so stay tuned for the next version of QUASSAR (QSR-200), that will be a more robust and smarter version.

For more information and videos, please click on the following links:

Google+
https://plus.google.com/+HABOTSSAR
YouTube
https://www.youtube.com/channel/UCNU4qTNbKUEccoNs_2CRuSg