Everyone, it seems, wants to save Hubble, but NASA’s preferred method, a robotic mission, might not be the best approach. (credit: NASA)

Robots and Hubble: a bad idea?

by Jeff Foust
Monday, November 1, 2004

In August NASA Administrator Sean O’Keefe paid a visit to the Goddard Space Flight Center to discuss the future of the Hubble Space Telescope. While the address was not promoted to the media in advance, its message soon leaked out to the press: NASA was moving ahead with planning a robotic servicing mission to the Hubble Space Telescope. “Everybody says, ‘We want to save the Hubble.’ Well, let’s go save the Hubble,” O’Keefe was quoted as saying. “Rather than just sitting there and talking about how we think we’re going to do it, we’ve got an option we’re ready to go with.”

That option, it appears, is to develop a robotic servicing mission to the telescope that is almost universally described as “ambitious”. While details about the mission are scant—largely because engineers are still figuring out what they need to carry out the work—it would involve state-of-the-art technology to replace key components on the spacecraft, notably its gyroscopes and batteries, and perhaps install new scientific instruments. The announcement is a notable turnaround for NASA, which early this year was ready to abandon the venerable telescope when it concluded that a planned shuttle servicing mission would be too risky in the post-Columbia environment. That decision was harshly criticized by scientists, politicians, and the general public, putting NASA on a course that led to O’Keefe’s speech at Goddard, as well as the initial award of contracts in late September to Lockheed Martin and MD Robotics to begin work on the robotic mission.

But is this the right course? A robotic servicing mission might do much to repair not just Hubble but NASA’s reputation as a can-do organization. It could also test new technologies that could play a role in the Vision for Space Exploration down the road. However, there are major hurdles that any such repair mission would face that will prove difficult—at the very least—to overcome. Moreover, by focusing on a robotic mission, NASA could be overlooking options that are more cost effective and carry better odds of success.

The challenges of cost and technology

There is widespread acknowledgment that Hubble is a valuable scientific tool that should remain in operation as long as it is feasible. With NASA no longer considering a shuttle servicing mission—a known quantity in terms of cost, schedule, and risk—the task of keeping Hubble alive falls to an as-yet undefined robotic mission. The viability of such a mission will depend on two key factors: the technical feasibility of one robotic spacecraft to serve another, and the cost to perform that mission.

A robotic mission to Hubble will face a large number of challenges, most of them rooted in the fact that the telescope was designed to be serviced only by the shuttle. The repair mission will have to be able to approach and dock with the telescope with little, if any, help from the telescope itself. The repair spacecraft will also likely require manipulator arms that can access the various equipment bays on the telescope to replace gyros, batteries, and scientific instruments—all of which were designed to be replaced by spacewalking astronauts. Much of the work will have to be done autonomously, or at least with limited control by engineers on the ground.

The first question naturally asked is if this is even possible with state-of-the-art technology? Given NASA’s reliance on the shuttle, the space agency has limited experience with autonomous docking. One mission, the Demonstration of Autonomous Rendezvous Technology (DART), scheduled for launch later this week, will test one method; XSS-11, an Air Force mission scheduled for launch early next year, will test similar techniques. DARPA’s Orbital Express mission, slated for launch in 2006, will test satellite docking and servicing, although with spacecraft specifically built for that purpose rather than an “uncooperative” satellite. Given the required schedule for a Hubble servicing mission, it’s not clear there would be enough time to incorporate the technologies and lessons from Orbital Express to a robotic Hubble repair mission.

An independent panel commissioned by the National Research Council (NRC) to study extending Hubble’s life raised those issues in an interim report released in mid-July. “The proposed Hubble robotic servicing mission involves a level of complexity, sophistication, and technology maturity that requires significant development, integration, and demonstration to reach flight readiness,” the panel noted in its report. “Based on information provided to the committee and the knowledge of members who have deep experience with shuttle flights and spacecraft servicing, the committee believes that the proposed robotic mission to Hubble will essentially be an experimental test program that is expected to accomplish specific programmatic objectives at the same time.”

That doesn’t mean that a robotic mission couldn’t be successfully carried out, but it does imply it carries significant risks. To compare a robotic servicing mission to other options, the Aerospace Corporation recently performed an “Analyses of Alternatives” (AoA) study for NASA, using cost, schedule, and risk as some of its metrics. The results, presented to the NRC panel during a hearing in August, were not encouraging to advocates of robotic repair missions.

The AoA study concluded that the probability of a successful repair mission was no better than 58%, and that was for a “Servicer Light” option that only extended the life of the spacecraft, with no replacement of scientific instruments, nor a deorbit module designed to avoid an uncontrolled reentry at the end of its life. A “Cadillac” option, which includes the deorbit module and the instrument replacement, was given just a 32% chance of success. By comparison, a shuttle servicing mission carried a 63% chance of success.

The robotic mission would not be cheap: $1.9-2.0 billion according to the AoA study, compared to NASA Goddard’s estimate of $1.3 billion. (The AoA study estimated the cost of a shuttle servicing mission at $2.2 billion, although some fraction of that money has likely already been spent in mission preparations prior to NASA’s January decision to cancel the SM4 mission.) Finding that money will be a serious challenge at an agency coping with the increasing cost of returning the shuttle to flight, as well as trying to fund an ambitious new exploration initiative. The Senate Appropriations Committee did include $300 million in NASA’s FY 2005 budget to pay for work on a Hubble repair mission. That money, though, came through the controversial use of an “emergency” spending measure that gets around spending caps; there’s no guarantee that the House, which included no such funding in its spending bill, will go along with the Senate.

The biggest issue facing a robotic servicing mission, however, is time. The two key subsystems on Hubble that a servicing mission needs to repair are its gyros and batteries. By 2007 or 2008 enough gyros will have failed on Hubble to render it useless for scientific work, although the spacecraft can be maintained in a safe mode until a repair mission arrives (as it was in 1999 before the SM3A shuttle servicing mission replaced failed gyros). However, by around 2009 engineers believe the spacecraft’s batteries will fail, rendering the spacecraft permanently inoperable since it won’t have the power to keep key systems functioning. While NASA Goddard believes that a robotic repair mission could be ready by the end of 2007, the AoA study estimated that a robotic mission would take between 57 and 65 months to build, pushing the launch into 2010 even if work started now. At that point, the study noted, there is only a 38% chance that the telescope will even be in a serviceable condition. In short, NASA could spend $2 billion on a robotic mission that has perhaps a 50-50 chance of success, only to have the telescope irretrievably fail before the mission can be launched.

page 2: two alternatives >>