Surveillance system with intelligent robotic surveillance device

A surveillance method, comprising:

• obtaining, by one or more computing devices, video data captured by one or more cameras;
• analyzing, by the one or more computing devices, the video data to determine whether there is any trigger event;
• determining, in response to a trigger event, a desired location based on the trigger event, and validating whether the trigger event is reliable;
• determining a selected robotic surveillance device among one or more robotic surveillance devices that is closest to the desired location;
• determining whether there is a security officer close to the selected robotic surveillance device;
• in response to determining that there is the security officer close to the selected robotic surveillance device, instructing the security officer to go to the desired location by riding the selected robotic surveillance device;
• and
• in response to determining that the trigger event is not reliable,
• determining whether the desired location is reachable by the selected robotic surveillance device, and
• directing, in response to determining that the desired location is reachable, the selected robotic surveillance to plan a path to reach the desired location and perform a responding action at the desired location,
• wherein the selected robotic surveillance device is further configured to track a person by a Multi-Scale Domain Adaptation Tracker (MSDAT) algorithm, wherein the MSDAT algorithm transfers features for image classification to a visual tracking domain via convolutional channel reduction, and
• the selected robotic surveillance device is configured to calculate a predicted location of the person indicated by the trigger event at a preset time, and move to and wait near the predicted location and perform the responding action at the preset time.

A robotic surveillance device, comprising:

• one or more processors;
• and
• a non-transitory memory storing instructions that, upon being executed by the one or more processors, cause the robotic device to perform operations, comprising:
• obtaining video data captured by one or more cameras;
• analyzing the video data to determine whether there is any trigger event;
• determining, in response to a trigger event, a desired location based on the trigger event, and validating whether the trigger event is reliable;
• determining a selected robotic surveillance device among one or more robotic surveillance devices that is closest to the desired location;
• determining whether there is a security officer close to the selected robotic surveillance device;
• in response to determining that there is the security officer close to the selected robotic surveillance device, instructing the security officer to go to the desired location by riding the selected robotic surveillance device;
• and
• in response to determining that the trigger event is not reliable,
• determining whether the desired location is reachable by the selected robotic surveillance device, and
• directing, in response to determining that the desired location is reachable, the selected robotic surveillance to plan a path to reach the desired location and perform a responding action at the desired location,
• wherein the selected robotic surveillance device is further configured to track a person by a Multi-Scale Domain Adaptation Tracker (MSDAT) algorithm, wherein the MSDAT algorithm transfers features for image classification to a visual tracking domain via convolutional channel reduction, and
• the selected robotic surveillance device is further configured to calculate a predicted location of the person indicated by the trigger event at a preset time, and move to and wait near the predicted location and perform the responding action at the preset time.