Professors teach robot to 'play ball'

Robotics and automation are getting a lot of coverage at the moment - more than I've seen before. I think it's another example of the accelerating pace of technology innovation. Automation and robotics has been researched for many years, but now a number of technologies are converging and the possibilities are multiplying. There is also a lot of research money available as automated and robotic systems are clearly seen as the way ahead. Especially in Mining where autonomous robotic mining machinery is seen as the best way to improve safety (by taking people out of hazardous environments) and production (by minimising process variability).

This article, Professors teach robot to 'play ball', talks about a robot that can field a ball. The article notes that its all about perception - how the robot can identify and intercept the ball - by itself. In this instance the robot is seeking a collision with the ball - but in industrial applications, the robots will be seeking to avoid collisions. Current research in mining systems tries to put transponders on all the individual things on a mine site - trucks, drills, people, buildings and use those to build up a dynamic real time model of the environment and let a clever system direct the traffic. Its way better than nothing, but if you aren't part of the model you can expect to get squashed.

Eventually, the model build needs to be independent of whether you have a transponder - it needs to be built visually (or by some analogue of vision).