Robonaut 2 Ready to Boldly Go Where No Humanoid Robot Has Gone Before

by Chris Gebhardt

Preparations are in full swing at the Kennedy Space Center for the final set of processing milestones leading up to STS-133. In particular, payload preparations are nearing their terminal phase as final cargo loads to the newly christened PMM Leonardo take place in the Space Station Processing Facility – including the final power-down, stowage, and installation of Robonaut 2 and its protective launch container into Leonardo.

Robonaut 2: The Next Generation of Orbital Interaction:

With just a little over two months left before launch, the story of Robonaut 2 is one that once again exemplifies NASA’s and its engineering partner’s ingenuity and ability to overcome seemingly insurmountable odds and time tables.

While development of Robonaut 2’s (R2’s) predecessors have been a part of the NASA engineering community since 1997, development of the R2 unit did not begin until 2007 under the Space Act Agreement which allowed General Motors (GM) and NASA to pool their resources and develop a successor to the Robonaut 1 series.

Through 2007, the R1 version of the Robonaut family had performed quite successfully in field and laboratory tests, proving the concept and validity of a dexterous robotic assistant. With R2’s development, GM and NASA created a faster, more dexterous, more technologically advanced humanoid robot than had ever been created before.

Nonetheless, while R2 was developed specifically for operational interaction with humans in space, no plans existed to actually launch an R2 unit into space until room opened up on one of the precious few remaining Space Shuttle missions.

“We weren’t expecting to fly,” states Dr. Ron Diftler, Robotic Systems Technology Branch, Automation, Robotics and Simulation Division at the Johnson Space System and lead engineer for R2. “Late last year we got the call that room had opened up on STS-133. It was decided by JSC engineering and Station management that launching R2 would be a good idea since its predecessors had already proven themselves reliable in lab testing.”

This decision meant taking the R2 unit through final construction, systems validation for flight, vibration testing, power compatibility testing, and final flight certification in eight months (December 2009 – August 2010).

Along the way, some consideration was given to exactly how R2 would be launched. While the preferred and final agreed upon method was always to launch R2 in a vibration absorbing box inside the PMM, Dr. Diftler confirms that it could have been possible to launch R2 like an astronaut – by strapping the unit onto a seat on Discovery’s middeck.

However, this idea was never seriously considered. “We talked about launching the unit like a human. But the protective box it’s launching in on the PMM is far better and provides substantial protection because the padded box minimizes the launch-induced vibrations.”

STS-133 Specific Articles:

In fact, the protective box is quite important to R2’s safe arrival on orbit since the unit was not designed to the Nth degree of vibration resistance. Therefore, placing the unit inside a vibration absorbing box significantly reduces the unit’s chance of being damaged by the launch vibrations of the shuttle.

“Putting it in its padded box reduces the changes of damage from launch vibrations and increases our changes of being fully operational when we unpack on Station,” notes Dr. Diftler. “That’s not to say we didn’t test it against vibration. It has passed all workmanship standards but we didn’t shake it as if it were launching on a hard pallet. The protective box limits our risks as much as possible.”

While the engineers are confident that R2 will make it to orbit undamaged, the dexterous, human-like robot is an experiment in and of itself and, as such, neither a significant maintenance plan nor large scale spare parts are being launched with the unit.

This makes the possibilities of effecting repairs on the unit limited. Recognizing this situation, future ISS crews have asked for the ability to “work under the skin” of Robonaut and the necessary training is being implemented.

However, while R2 will soar to orbit with Discovery and officially arrive on the ISS with the berthing of PMM Leonardo, the unit itself will not be unpacked immediately. Situated in the “rear” portion of the PMM, R2’s unpacking will not begin until January 2011 at the earliest based on current ISS crew and visiting vehicle schedules.

“Because of the other payloads going up with R2 it probably won’t come out of the PMM until January. And even then we’ll only have a very limited amount of crew time to work with R2 – somewhere on the order of 2-6 hours per month. So it’s going to take a while to get through the opening assessments and tasks,” notes Dr. Diftler.

Once initial unpacking is complete, R2 will spend most of its time “just sitting” as the ground teams run through their series of evaluations on the unit and its various sensors.

As Dr. Diftler relates, “We’ll be going through a rather challenging series of events from launch to integration on the station. Once initial checkouts are complete we will slowly start moving Robonaut – just the joints in single joint mode and moving up to simple demonstrations.”

At first, R2’s movements will be restricted relative to its large range of arm motion. Once the initial checkouts are complete, R2 will be fastened to a work stand in the Destiny science lab and placed in front of its dedicated task board – which will help further validate and fine-tune the Robonaut’s systems for the micro-gravity environment.

In essence, R2 will not interact with any critical Station hardware or systems; it will only interact with non-critical hardware that can be used to verify the unit’s operations. During this phase, the ISS crew will have some interaction with unit to make sure it can respond properly to crew commands – among them an emergency shut-down procedure initiated by the steady application of light force by an astronaut to one of R2’s forearms.

This emergency shut-down procedure was developed as a quick way to deactivate Robonaut for crew safety – though it is currently thought that this procedure would only be used in the event that an inadvertent commend was sent to the unit by a crewmember. Nonetheless, the shut-down procedure is there should the crew need to use it.

After this initial series of orbital tests, the engineering team hopes to command R2 through a series of basic tasks in cooperation with the ISS crew. These tasks could include, among other things, vacuuming air filters and holding tools/items for the ISS crew and they work on nearby systems or science experiments.

As humorously stated by one of R2’s engineers: “R2 is capable of holding tools for people for hours on end without complaining. And it only has to be told once to hold a flashlight on just the right spot.”

But contrary to what Science-Fiction would have us believe, and despite all the possible tasks R2 could perform over its lifetime on the ISS, R2 has what its engineers refer to as limited intelligence.

“Right now it just has the programming to perform basic dexterous tasks. Right now we’re having it learn to identify certain objects, like fasteners, so that we can give it a pre-programmed task,” states Dr. Diftler.

The identification of certain objects, like fasteners, in combination with his sensory input, will allow R2 to have a limited perception of its surroundings. This will allow the robot a very limited degree of autonomy during operation.

As Dr. Diftler relates, “A crewmember can access a pre-programmed tasks for R2 and then place the unit in front of a specific area of the station’s interior. R2 can then use its sensors to scan the environment and identify a fastener from its memory. But it won’t know how far away it is from the fasteners.

“By combining its memory of what a fastener looks like and information from its sensors R2 can identify where the fasteners are and then adjust its movements accordingly.

“So initially R2 will run on a series of pre-programmed scripts that it can then adapt based on sensory data. It doesn’t and won’t think for itself but can rather use sensory data to adapt its pre-programmed tasks.”

Later in its operational life, new capabilities for R2, such as vision, can first be tested on the ground on an identical R2 unit and then transmitted up the Station for implementation on the Orbital R2 unit.

And vision might just be the first in a line of potential upgrades and enhancements for R2 units. In fact, limited intuitive thought could be a possible addition.

“We can teach R2 a higher level,” notes Dr. Diftler. “We can teach it ‘go find the box.’ It will search its memory, identify what a box looks like, and then scan its environment and find the box. And in that time it will only care about finding that box. It won’t have anything else on its mind.”

In all, R2 represents and amazing leap forward in the field of robotics – dexterous, humanoid robots in particular.

“It’s very exciting to go from a lab, to interacting with the crew, and now taking the significant step of getting it to orbit,” notes Dr. Diftler.

“Robonaut, being a tool for crews, has always been evaluated against the various environments crews will be going to. As NASA’s plans have evolved, we’ve tried to keep Robonaut relevant. And we’ll continue to do that.”

Related Articles