Delayed feedback, oscillations and the police

This story from Yahoo is hilarious – about a policeman who chased himself for 20 minutes. In a nutshell, a CCTV operator sees a man acting suspiciously, unaware that it is in fact a plain clothes policeman. He radios through to the police, who pass the message on to their nearest policeman to the scene, who is, of course, the man himself. And off he goes, to find this suspiscious man, unaware that it’s himself, guided by messages relayed from the CCTV operator.

WIth a bit of physics experience, I can see how the chase would unfold. The officer is in Smith Street, say, but he gets a call on the radio to go to nearbyJones Street. That’s because there’s been a bit of a delay getting the message to him – Jones Street was where he was two minutes ago. When he gets to Jones Street, and sees nothing, he gets told that, actually, the man is in Smith Street – again, where he was two minutes ago. Back to Smith Street he goes, to be told that he should be in Jones Street, and so it continues.

I don’t know that’s exactly how it went, but it’s plausible. The thing that’s controlling the behaviour here is the fact that the feedback is delayed. When this happens it can often lead to oscillation (e.g. Smith to Jones to Smith to Jones…) It’s something that has to be thought about in control systems of aircraft, for example. If the aerilons respond too slowly and too severely to small fluctuations in an aircraft’s roll, the aircraft can start oscillating. Not a good situation to be in. There’s ways of analyzing just how stable a system will be when it receives feedback, and it’s a key analysis to be done in many situations. The CCTV one is a comical one of little significance, but, in many modern systems full of electronics, the systems engineer needs to have done his or her job properly or you could be in trouble. 

Leave a Reply