A voice controlled or command controlled robot is shown, with Picamera for surveillance of remote areas like mines, borders, jails etc. It's features include face detection, face recognition, live streaming to multi devices and option of either voice control or input typed control.
Working:
It uses MQTT protocol to send commands over the internet from either a laptop or phone to the Raspberry Pi on the robot. The robot interprets the message sent through the running Python code and accordingly operates the motors connected in the desired direction. The robot can be controlled from anywhere in the world through the internet and can also stream the data from the picamera attached to it to multiple devices in real time. In the video, 2 laptops are shown, one at the extreme left showing the output in the command window showing the movement of the bot. The laptop at the right has a application called "MQTT Box" where the inputs like "Forward", "Reverse", "Right" or "Left" are typed for the bot to move in the specified direction. Additionally, the commands can be given by voice through an android application(not shown) called "MQTT Dash" where voice to text is deployed to convert the voice command to text to send to the robot. 

In this video, the face detection part of the system is shown. When a faces are detected, it will show the number of faces detected in the frame. This can be useful in surveillance systems, along with face recognition capabilities(not shown) to recognize and detect people in remote areas like mines and borders, or jails to increase efficacy.  Additional details of the project can be viewed in the below link published in the IJAST International Journal   http://sersc.org/journals/index.php/IJAST/article/view/21218/10752