How Robonauts Work

I, Robonaut

Robotic probes and rovers have been traveling to Mars since before we landed on the moon. In 1965, Mariner IV sent back the first close-range images of the red planet. In 1997, the Pathfinder rover provided unprecedented detail on the Martian atmosphere and surface. Moreover, who can forget the remarkable contributions of Spirit and Opportunity, the two Mars rovers launched in summer of 2003 that so outlasted their original mission?

NASA has based its robotic astronauts on a humanoid design. The first of these, Robonaut 1 (aka R1), featured a head, two eyes, two arms and two five-digit hands. Designers protected R1's head with an epoxy-resin helmet and mounted the head on a jointed neck, which permitted it to turn from side to side and to look up and down. Within the pioneering robonaut, two video cameras delivered stereovision to the operator and enabled R1 to track objects. Stereovision mimics human vision by comparing images from a right and left "eye" (camera) and using parallax -- the apparent difference in an object's position caused by the different viewing angle of each eye -- to determine depth and detect motion. R1's arms were capable of a greater range of motion than human arms and packed more than 150 sensors each.

NASA began construction of R1 in 1997, and it served as an experimental platform in laboratory and field tests until 2006. It was a successful proof of concept, but it never left the lab.

In 2006, NASA signed an agreement with General Motors to produce Robonaut 2 (R2). GM was also developing dexterous robots at the time and had worked with NASA on the lunar rover. NASA unveiled R2 in February 2010, and the robonaut traveled to its permanent home on the International Space Station Feb. 24, 2011, on one of the final space shuttle missions.

Like R1, R2 is designed to help humans and to automate repetitious, dull or tiring tasks -- such as setting up the tools and equipment necessary for missions -- freeing astronauts to concentrate on tasks that only they are qualified to perform.

Think of R2 as R1-plus -- smaller, cheaper, more advanced and capable of surviving the rigors of launch and of space. R2 delivers more than 350 sensors, 40 of which it uses to detect its surroundings. That includes four visible-light cameras in its eyes and a fifth infrared camera in its mouth to aid in depth perception. Its stomach contains 38 computer processors. Although its strength is on par with R1's -- it can lift around 20 pounds (9 kilograms) -- R2 is more adept with its handy appendages: Whereas R1's hands were akin to an astronaut's gloved hands, R2's are more like ungloved human hands.

R2 can manipulate a blanket, pick up an envelope and grip a dumbbell, but its dexterity is greater than the sum of its parts. Users can control R2's joint stiffness, which gives R2 a leg up over typical "positionally controlled" robots like automobile assembly robots, which lack "give" in their systems and must line up perfectly to do their job. Such a robot would be lousy at putting a peg in a hole; even a slight misalignment would cause it to smash the peg into the area around the hole. R2, conversely, can "feel" its way home, moving the peg softly forward and making small, sliding corrections if misaligned, like a human would. R2's flexibility also makes it safer for its human companions, who can stop its motion without much force, thereby avoiding injury.

Here are the specifications for Robonaut 1 and 2:

Specifications Robonaut 1
Robonaut 2
Height6.23 feet (1.9 meters)3.33 feet (1.0 meters) (waist to head)
Weight410 pounds (182 kilograms)330 pounds (150 kilograms)
Structural Materials
Mostly aluminum with Kevlar and Teflon padding to protect it from fire and debrisPrimarily aluminum with steel, nickel-plated carbon fiber and nonmetallics
Computing Platform
PowerPC processor38 PowerPC processors
Operating System