« Back

Algorithms Help Household Robots Perform Tasks

Students from MIT’s Computer Science and Artificial Intelligence Laboratory have designed algorithms that help household robots pick up and place items and complete various chores. Most experimental robots use motion-planning algorithms to help them avoid collisions, navigate, and perform tasks, but these algorithms must search among a great number of possible movement options. Typically, these types of algorithms help the robot identify those movement paths that avoid collisions. One of the new MIT algorithms looks for collisions, but also enables the robot to avoid bad path selections when moving an object. It allows the robot to place the object into an optimal, desired movement path. Users can also modify the algorithms via an interface that lets them input new parameters describing new types of objects with which the robot may interact, which should enable the robot to smoothly pick up and place a wide variety of objects in a specific manner rather that using a universal algorithm for the task. The researchers, who used the Willow Garage PR2 robot in their experiments, will present their work at the IEEE International Conference on Robotics and Automation in May. (PhysOrg)(MIT News Office)
 

Comments
Trackback URL: