We undertook a tricky task – create a security robot that can do its job, on its own, for a full eight hour shift. Of course, the vast majority of those security shifts are boring and routine – in fact deadly dull. That is what makes them hard for a person to do well, and ideal for a machine. But security is really needed when things aren’t routine, when things stop being boring, when things go wrong. That is when you have to have intelligent, correct responses to the incident, and you have to have them now. Unfortunately, that kind of intelligent, situationally-aware response is exactly what machines are not good at.
Unless they have been designed from the ground up to provide that kind of response. To respond appropriately to an incident there are four key capabilities(1):
- The ability to detect a potential problem – Observe,
- The ability to recognize what is going on – Orient,
- The ability to determine the ‘appropriate’ response, and – Decide
- The ability to execute the response -Act.
These four capabilities must be resident in the robot, it can’t rely on having a human doing all the work, otherwise it doesn’t provide the force multiplier effect that is needed. After all, if you have to keep a person around the ‘drive’ the robot, you haven’t gained, you’ve actually lost. Instead of having a security officer doing his or her rounds, you now need a highly trained (and highly paid) robot technician on staff 24/7 just in case the robot has to deal with a security incident.
Let’s set the scene. A shopping mall in the middle of the night. Things have been quiet, as they usually are. About four hours ago, the security officer told the robot to run the normal night patrols for its entire eight hour shift. After that the officer was free to attend to other duties.
Now, if we were running an assembly robot in an industrial setting, that would be it. In that kind of tightly controlled environment all that is needed is a simple, fixed program because nothing is allowed to ‘disrupt’ the robots. But in a human environment, you can’t have that level of control. Unlike a robot, people never do the same thing twice in the same way.
So early in the evening the robot detected that a mop bucket had been left in a service corridor, and four things happened: The robot detected an obstacle, determined that it was a typical item, figured out how to get around the obstacle, and then executed the appropriate behavior. All in a nights work. Later, things get more complicated – and that is why the robot needs to be intelligent.
While the robot was on patrol, the security officer got a door alarm from near the loading dock. This could be nothing, or it could be trouble. She dispatched a command to the robot – “Stop the patrol and get to the East Loading Dock.” The robot now has to figure out the best way to get from where it is right now to the loading dock, and factor in the information learned during the shift. The most direct route would use the service corridor that is blocked by the mop bucket, so come up with the best clear path, and get moving!
At every moment the robot is in the loop: planning the best action, predicting the outcome, executing the action, comparing the results with the prediction, and then planning again. This allows the system to respond appropriately to the situation at all times. If it encounters another obstacle on the way to the loading dock, it has to know more than the fact that there is an obstacle. If this a box int the way, the robot just goes around it. But if this obstacle is an intruder the robot needs to behave completely differently, turning on strobes and sirens and sending an alert to the command center. There is no ‘one size fits all’ response – the correct action must be determined from the circumstances – just like a well trained human would do.
If a robot is going to work with people, and if people with critical jobs are going to rely on that robot, the robot needs to be intelligent and autonomous to hold up its part of the bargain.
Where is your robot?® Ours are made in the USA and are distributed by Vigilant Robots.