3Qs: Controlling a robot from another planet

August 21, 2012

Marsette Vona, an assistant professor in the College of Computer and Information Science, explains the challenges of controlling a Martian rover.

Prior to joining the North­eastern fac­ulty as an assis­tant pro­fessor in the Col­lege of Com­puter and Infor­ma­tion Sci­ence, Marsette Vona worked for NASA’s Jet Propul­sion Lab­o­ra­tory as part of the teams that put both the Spirit and Oppor­tu­nity rovers on the sur­face of Mars. We asked Vona, who is devel­oping robots that can detect uncer­tainty in their envi­ron­ment, to explain how Curiosity, the newest and most advanced rover on the red planet, and its crew on Earth handle the logis­tical chal­lenges of space exploration.

The vast distance between Earth and Mars makes it impossible for NASA scientists to communicate or control Curiosity without a time delay. To compensate for this time delay, what did engineers and programmers need to consider when designing the rover and planning the mission?

The Earth to Mars radio delay, which ranges from about three to 20 min­utes, adds sig­nif­i­cant com­plexity to robotic mis­sions on Mars. Another chal­lenge is the logis­tics of sched­uling the Deep Space Net­work radio dishes, which are shared with other dis­tant space mis­sions and which are our only means to com­mu­ni­cate at inter­plan­e­tary dis­tances. Yet another issue is that each Mars day is about 40 min­utes longer than an Earth day, making it dif­fi­cult to syn­chro­nize human oper­a­tors living on Earth time with rover activity in day­light on Mars.

So in order to get much done, the robots we send to Mars must be fairly intel­li­gent: They must be able to operate autonomously for some amount of time. We cannot “remote con­trol” them as we can do, for example, with sub-sea explo­ration robots on Earth. Instead, we typ­i­cally send them com­mands before dawn on each Mar­tian day. They exe­cute those com­mands autonomously, pos­sibly making some deci­sions on their own, and radio back results and status infor­ma­tion during the fol­lowing Mar­tian night.

Your expertise lies in understanding how robots handle uncertainty, such as uneven terrain or unexpected surroundings. How does Curiosity handle similar challenges and how can those technical advances be used on Earth?

Curiosity uses a com­bi­na­tion of tech­niques to reli­ably move around in the sand and rock envi­ron­ment of the Mar­tian sur­face. First, its mobility system, com­posed of six wheels and a sus­pen­sion called a rocker-bogey, enable it to roll right over rocks up to 50 cen­time­ters (about 20 inches) tall. Second, it will auto­mat­i­cally attempt to drive around larger obsta­cles, which it looks for using com­puter vision algo­rithms in live video feeds from front– and rear-facing cam­eras called “haz­cams.” Third, it can use other com­puter vision algo­rithms to ana­lyze longer-distance images from mast-mounted “nav­cams” for sev­eral pur­poses, including “visual odom­etry” to esti­mate how far it has actu­ally trav­eled (which may differ from wheel rota­tion data because of slip­page in the sand), and target tracking to mon­itor progress toward a vis­ible goal such as an inter­esting rock.

A mobility system like that on Curiosity could be used on Earth for robots that must travel over rubble after, say, an earth­quake or other dis­aster. Its com­puter vision algo­rithms for autonomous nav­i­ga­tion and obstacle avoid­ance are also extremely useful; related sys­tems are used on Earth for autonomous cars. Some of my own cur­rent research focuses on adapting algo­rithms like this for walking robots, which could help humans travel over very uneven ground, or which could even­tu­ally replace us in under­taking haz­ardous tasks.

How is Curiosity’s ability to explore Mars different from the Spirit and Opportunity rovers?

Over the last 15 years we have suc­cess­fully landed four rovers on Mars. New tech­nolo­gies suc­cess­fully tested in each mis­sion are incor­po­rated into later ones. Also, each mis­sion has approx­i­mately dou­bled the rover size, which is impor­tant because larger rovers can carry more sci­en­tific instru­ments and can travel far­ther and faster.

Spirit and Oppor­tu­nity tested pre­cur­sors to the auto­matic nav­i­ga­tion, obstacle avoid­ance and visual odom­etry algo­rithms on Curiosity. Curiosity, how­ever, is twice their size (about 3 meters long) and has addi­tional autonomous nav­i­ga­tion capa­bil­i­ties that should enable it to travel far­ther on its own. Curiosity also used a new “sky crane” landing system instead of airbags, and it is pow­ered by a radioiso­tope thermal gen­er­ator, unlike Sojourner, Spirit, and Oppor­tu­nity, which were solar pow­ered. This should enable it to operate over a full Mar­tian year, whereas the pre­vious mis­sions were typ­i­cally only active in the plen­tiful sun­light of the Mar­tian summer. Finally, and pos­sibly most sig­nif­i­cantly, Curiosity car­ries sev­eral new instru­ments that will be used to ana­lyze sam­ples of the Mar­tian rock to detect water, carbon com­pounds and other bio­log­i­cally impor­tant mate­rials. This will both help us under­stand whether life ever existed on Mars and also the extent to which we will be able to use mate­rials on Mars to help sup­port future human explorers.

Photo by Mary Knox Merrill