Researchers at MIT have produced a new algorithm that enables sensor laden robots to focus on parts of the environment that change more frequently than others without losing track of the regions that change slowly.
The unveiling will take place at the Institute of Electrical and Electronics Engineers’ International Conference on Robotics and Automation in May.
The MIT researchers will also present a paper on the test run of the same algorithm on undersea sensors which the University of Southern California researchers are using to study algae blooms. The sensor algorithm is the work of Daniela Rus, Mac Schwager and Stephen Smith.
It is designed for use in robots that are monitoring a specific environment for long time periods. Since the data that they collect while tracing the same routes repeatedly varies at different rates in different parts of the environment the algorithm will help make the robots work easier by adjusting the velocities of the robots as they work.
In a quest to optimize the acquired knowledge the sensor algorithm will work out just how often and how soon the robot needs to make a repeated journey down the environment path that it is monitoring. The researchers involved in the project have called it a monstrously complex calculation.