A data lake architecture for telepresence robots
Telepresence robotic avatars one such as from Avatar, where a operator can control remotely located robot through a jacket. There are two different kinds of data generated here:
-
Sensory Data and:
Since the nature of these sensors generate time dependent data. The information generated from jacket consists of sensors such as following:
- Feedback
- Galvanic Skin Receptors
- Haptic Sensors and
- Motion Sensors
-
Video information:
The video and audio feedback from robot while the operating the robot.
Considerations and understanding on what basis we've choosen this architecture, you may check out this link.
Following are technical stack we've used:
# Automate shell script is responsible for getting up Kafka cluster up,
# running Kafka Producer and Consumer, and
# Connecting PySpark to MongoDB
$ bash automate.sh
# Change script is responsible for publishing the data to HDFS through URI
$ bash change.sh
Most of our research work, and paper to the conference can be found at this link