There could be deep learning methods and processes / algorithms created to study the process of catching criminals, and there could be systems that automate the entire process of policing. It would go far beyond facial recognition and robots. Obviously, it’s a lot more complicated than combining a human with cybernetics. You need the hardware and software that runs on the robotics to make it possible to catch criminals, and to solve cases.
There would be entire deep learning mechanisms that would study what different individuals in Police Departments do, and integrate their interpersonal relationships in how they help each other to solve the crimes in their various titles or positions. The program would then compile all known data from public sources, such as from surveillance, and accurately identify suspects, that would be automatically produced for human police officers, that could then check the accuracy of the computer systems.
There would need to be a lot of security on the computer systems. Obviously, criminals would try to compromise the systems’ integrity, and either make the systems go down such as with Denial of Service attacks, or compromise its integrity – or make it catch the wrong people.
These systems are possible through the integration of both local, state, and Federal law enforcement and domestic surveillance resources, such as pooled data that is gathered by the National Security Agency. If there is evidence of someone committing a crime, it should be provided to law enforcement systems that can keep the integrity of the information, while protecting the public from its disclosure.
The system could also employ calculations, such as situational awareness for police and riot squads, such as automatic identification and tracking of specific individuals with Heads Up Displays (HUD) in their helmets or shields.