Thursday, August 11, 2011

On event processing and robotics

In my standard presentation about proactive event-driven computing  (I gave one of those in Almaden earlier this week),  I always show these two pictures:   I am about to spill coffee on my laptop, and my robot detects it and grasp the cup.    This is the simplest example to explain the idea and people understand it and typically smiling (or laughing),  in one instance one commented that it is not cost-effective to use robot for these purposes, which is probably true, but he missed the point of what I've tried to do.
BTW -- spilling coffee is something I do always on a daily basis (today I spilled ice-coffee), I am very talented in spilling coffee on everything (though until now not on a laptop, I have a rule of not drinking the coffee too close to the laptop), so having such a robot would be extremely useful to me!   

In the EPTS meeting in Trento we heard a talk by Sebastian Wrede about event-based robotics,  but I admit that I have never followed up with the robotics guys.    

Today in AAAI there has been a keynote talk by Kurt Konolige about robotics,  after the talk I came forward and introduced myself by saying that the closes I came to robotics was reading Asimov books (which  probably was a bad starting-point for a conversation with him), and explained him a little bit about what event processing is and then moved to the coffee spilling scenario.  His reaction was that in the robotics world they deal which more more basic stuff:   How to improve the vision and perception, and how to calibrate the robot's arm to be more accurate, so he thinks that the speed in which the robot will identify that I am going to spill coffee and the accuracy in which he can grasp the cup are not in the level that the current robotic technology can do this scenario -- which does not mean that I'll stop using it!   His message is that before robots will be able to participate in "high level" tasks, as he called what I described to him, there are many "low level" problems to be solved.      Well - I'll have to wait patiently,  or find some other pointer in the robotic community who will be interested in the synergy between robotics and event processing -- I should definitely start active search.  

No comments: