TR Editors' blog

Modeling Sneaky Robots

Researchers create a stealth-mode behavior

Kristina Grifantini 05/20/2009

  • 4 Comments

Imagine a guard robot trying to apprehend an intruder or catch an escaped prisoner. Instead of simply tracking the target, ideally the robot would know how to creep up stealthily, by crouching behind objects and hiding in the shadows. This way, it could surprise and apprehend its target before it can get away.

Creepy as this might sound, a team at Seoul National University in Korea has created an algorithm that models just this sort of stealthy behavior. Professor of engineering Beom Lee and first author Jung-Hee Park designed simulations in which a robot waits in the shadows and moves quickly between obstacles to intercept a target.

The simulations take into account the size of the robot, the size of objects it can hide behind and the speeds of moving target. In the videos below, the robot, blue, tracks the invader, red. The blue line marks the robot's planned path and its projected interception point. In this movie, the system is in "stealthy mode", moving only when the intruder won't see it:

In this movie, the system moves with only the "intercept intruder" program, leaving itself visible to the intruder:

Video

It's a pretty neat idea, though robots will have to be much less noisy before being truly stealthy.

Print

Close Comments

To comment, please sign in or register

Forgot my password

jschuman

5 Comments

  • 993 Days Ago
  • 05/21/2009

guard bot

Reply

mkogrady

423 Comments

  • 993 Days Ago
  • 05/21/2009

Re: guard bot

Build a Tazer into the thing JUST to make sure it shocks the intruder.

Reply

hankjmatt

5 Comments

  • 833 Days Ago
  • 10/28/2009

The shock alone might just incapacitate the intruder..

Reply

Shiladie

56 Comments

  • 771 Days Ago
  • 12/29/2009

Sensor reliant

Very neat tech, but without the sensor pack to know where the target's sight lines are, this is relegated to simulation only.

The movies here don't really show a very complex scenario for their script to truely shine.  I would really like to see a human controlled red dot moving around the maze in an unpredictable manner, and see how the script responds to it.  piviotally, whether the script will anticipate where sight lines will soon form with current target direction and speed, and move out of those areas.

Reply

About

Insights, opinions, and our editors' analysis of the latest in emerging technologies.

Subscribe to the TR Editors' blog RSS Feed

Advertisement
Advertisement

Facebook

Advertisement