Global Technology - July 2016

Comments Off on Global Technology - July 2016
Global Technology - July 2016 LoadingADD TO FAVORITES

Once they’ve mastered the skills of “toddlerhood,” humans are pretty good at what roboticists call “motion planning,” which might involve the thinking needed to slip their hands around a screen to connect an unseen cable.  For robots with multi-jointed arms, motion planning is a hard problem that requires time-consuming computation.

Consider this.  When you think about a car assembly line, the entire environment is carefully controlled so that the robots can blindly repeat the same movements over and over again.  The car parts are in exactly the same place every time, and the robots are contained within cages so that humans don’t wander past.  But if your robot is using motion planning in real time and a part is in a different place, that requires a lot more processing

At the recent Robotics: Science and Systems conference in Ann Arbor, Michigan, Duke University researchers introduced a specially-designed computer processor for motion planning that can plan up to 10,000 times faster than existing approaches while consuming a small fraction of the power.  The new processor is fast enough to plan and operate in real time, and power-efficient enough to be used in large-scale manufacturing environments with thousands of robots.

Why is this important?  Speedy motion planning saves the time and expense of engineering the environment around the robot.

Motion planning has been studied for thirty years, and recent advances have brought the time required to find a plan for a sophisticated robot down to a few seconds.  With few exceptions, these existing approaches rely on general-purpose CPUs or computationally faster but more power-hungry graphics processors (GPUs). 

But while a general-purpose CPU is good at many tasks, it cannot compete with a processor specially designed for just a single task.  The Duke team designed a new processor specifically for motion planning.

The technology works by breaking down the arm’s operating space into thousands of 3D volumes called voxels.  The algorithm then determines whether or not an object is present in one of the voxels contained within pre-programmed motion paths. Thanks to the specially designed hardware, the technology can check thousands of motion...

To continue reading, become a paid subscriber for full access.
Already a Business Briefings subscriber? Login for full access now.

Subscribe for as low as $135/year

  • Get 12 months of Business Briefings that will impact your business and your life
  • Gain access to the entire Business Briefings Research Library
  • Optional Business Briefings monthly CDs in addition to your On-Line access
  • If you do not like what you see, you can cancel anytime and receive a 100% pro-rata refund