|
|
|
Thread Tools | Display Modes |
#1
|
|||
|
|||
Slip Sliding Away (Mars Rovers)
http://marsrovers.jpl.nasa.gov/spotl..._20040309.html
Slip Sliding Away Jet Propulsion Laboratory March 09, 2004 On February 9, the wake up song for Opportunity was 'Slip Sliding Away' by Simon and Garfunkel in honor of the experience the prior day when Opportunity first tried to scoot up the crater slope at Meridiani Planum. The ambitious rover aimed for its target and willingly went forth up the slope; nevertheless, Opportunity slipped and slid in the sand, making it only half way to its target. But ignorance is bliss, and while Opportunity's wheels had a hard time gaining traction on the sandy surface, the rover's brain (or computer) had a hard time grasping that it hadn't successfully made it to its target. The little rover didn't have to encourage itself to make it up the crater slope later by chanting, "I think I can, I think I can," because the little rover thought it actually did make it up the slope the first time. Measuring the Distance Traveled on Mars "Like a car on Earth, each Mars rover uses an odometer to click off the distance its wheels travel to measure and register how far the vehicle has moved," explains Randy Lindemann, rover mobility lead. One revolution of the rover wheel equals 80 centimeters or 2.6 feet, so after the wheels have revolved four times, the rover believes that it has moved forward 320 centimeters or 10.5 feet (80 centimeters X 4 = 320 centimeters). Alas, Opportunity's wheels had a hard time grasping onto the sandy ground around the crater and the wheels spun in place before they actually gained tracking. "As Opportunity's wheels turned and ticked off 80 centimeters (2.6 feet) each revolution, they eventually spun four times, calculating to what it thought was a distance of 320 centimeters (10.5 feet). Thus, Opportunity believed it had reached its goal, when in reality, it had spun in place 50 percent of the revolutions and only really made it 160 centimeters (5.25 feet)," said Lindemann. Outwitting the Odometer In order to prevent any future missed targets, Opportunity's mobility experts quickly started trying to predict exactly how far the rover would slip down a slope or fall short of a target while climbing up a slope due to the loose terrain along the steep angles of the crater wall. "Since the rover isn't on cruise control and can't rev its engine to get some extra oompf to go up a hill, we continually have to outwit Opportunity's odometer and command the rover to go farther or shorter than the real target distance," said Rover Driver Eric Baumgartner. Rover Drivers and their Crystal Ball "Well, we don't quite have a crystal ball to predict slippage of the rover wheels," said Lindemann, "but our team of mobility experts and rover drivers can make predictions using a rather nice piece of paper with a curved line on a standard plot." How well does this prediction on paper work? It's all about prior testing. While Opportunity was cruising through deep space on the way to Mars, engineers on Earth tested the rover's mobility using an engineering model of the rover of the same weight and size with identical wheels. Engineers affectionately named it the "SSTB-lite rover." SSTB-lite stands for Surface System TestBed, and the lite means that this rover doesn't have any of the appendages, such as the robotic arm, high-gain antenna, or panoramic camera mast assembly. "Our sandbox was a variable-tilt platform covered with 6 inches of dry, loose sand similar to what is used with construction cement. Throughout the weeklong test, we plotted how much the test rover slipped at different angles going up, down and across the simulated martian terrain," explained Lindemann. "We never expected the incredible result that the rover's slipping behavior at the Meridiani site would be almost identical to its behavior on dry and loose sand. The reason that this surprised us so much is that the soil at Meridiani is nothing like dry beach sand from the Earth in terms of the minerals that it is made of or even how it was formed. What we have learned is that the primary characteristic of any loose soil in terms of how the rover will drive on it is determined by the characteristics of the friction between all of the tiny grains -- pretty much no matter what they are made of," said Lindemann. How Far Will Rovers Slip and Slide? The test created a set of trustworthy mobility plots. The plots show rover drivers that, at a 15-degree angle facing down, the rover will slide an extra 25 centimeters downward for every meter it is trying to go. The chart is not a straight, even line. At a 20-degree angle downward, the rover will slide 55 centimeters (1.8 feet) down, whereas at 20-degree angle upward, the rover will slip 90 centimeters (2.9 feet) in place and only move forward 10 centimeters (3.9 inches) out of a drive of 100 centimeters (3.3 feet). "For the type of sand Opportunity is in, the dead-end point where the rover simply can't climb upwards anymore, regardless of how many times the wheels turn, is 25 degrees," explained Baumgartner. "We've already gone up slopes as steep as 22 degrees on Opportunity Ledge. Luckily, the Long Term Planning team has found places around the crater with slopes lower than 25 degrees, so we can eventually get out of this crater where we landed," said Baumgartner. Driving Spirit Versus Opportunity Baumgartner is one of eight rover drivers. There are four for Spirit and four for Opportunity. Two rover drivers are on station for any given sol. "It's a lot of fun to come to work and drive the rovers on Mars and drive the rover arm, which is a whole other story in itself," said Baumgartner. The rover drivers for the two rovers have had very different experiences due to the difference in terrain at the two sites. "At the Opportunity site, it's like we're driving on snow with little traction on the slopes, but the Spirit site is like monster truck driving with the rocky terrain," laughed Baumgartner. Spirit is on flat, rocky terrain, while Opportunity is in steep terrain with a low distribution of rocks, so the two rover driving teams will be swapping lessons learned when each rover moves to a different location. "If Spirit makes it to Bonneville crater, the rover drivers will have to rely on what we've learned at Meridiani, and once we get out of the crater at Meridiani, we may need to learn how to drive far in a rock-strewn landscape from the Spirit team," said Baumgartner. On Your Mark, Get Set, Rove! In order to build the drive commands for Spirit and Opportunity, rover drivers use sophisticated software created at NASA's Jet Propulsion Laboratory to make a three-dimension simulation of the martian terrain using navigation camera images and a simulation of the actual movement of the rover. Rover drivers must simulate the rover movement in a computer program because there isn't enough time to sprint to a test bed (a sandbox with an engineering rover model) every day to run the command sequence. Schedules are extremely tight to turn around new, complex robot moves and send the commands to a planet over 150 million miles away. "We run many, many simulations -- up to 20 per move -- to figure out what the best path is for each segment of the drive," explains Baumgartner. Rover drivers for Opportunity have to add in the slippage estimates created by the mobility experts. Rover drivers use the slope information and tell the rover to stop short of its target when it's heading down a slope so that the rover will slide into home plate on its own. When the rover has to trek up a slope, engineers do the opposite since it is harder for the wheels to churn upward. If the rover needs to go up a slope, engineers command the rover to actually go farther than necessary to compensate for the upward tilt. "I didn't think adding in slippage would be part of my work as a rover driver," said Baumgartner. "The foresight of mechanical team was tremendous, and if we didn't have the slippage chart, we would be putting our thumbs to the air and saying, 'I think the rover will slip about this much here,'" laughed Baumgartner. "Within about five hours every day, we have to write hundreds of lines of commands to drive the rover to the scientists' dream location. Without the efficient slip estimates, which the mobility team constantly refines, we would be in a lot of trouble here," said Baumgartner. Body Movin? On February 18, the wake up song for Opportunity was "Body Movin" by the Beastie Boys in honor of the 15-meter (49-foot) drive, which was Opportunity's farthest distance to date. "We were off by less than 0.5 meters (1.6 feet) from our desired final destination, which is an error of about 3%. In contrast, Opportunity actually slipped between 10% and 17%, so without this slippage planning, we would have been off by as much as 2.6 meters (8.5 feet). That would have cost us another day on Mars to get close enough for the pancam and mini-TES work that was planned at the target," said Lindemann. "We've moved from sad folk music to body movin' rap music for wake up songs, so I'm pretty happy," said Lindemann with a smile. |
#3
|
|||
|
|||
Slip Sliding Away (Mars Rovers)
On 10 Mar 2004, Ron wrote:
http://marsrovers.jpl.nasa.gov/spotl..._20040309.html Slip Sliding Away Jet Propulsion Laboratory March 09, 2004 On February 9, the wake up song for Opportunity was 'Slip Sliding Away' by Simon and Garfunkel in honor of the Slip Sliding Away was done by Paul Simon after he went solo. |
#4
|
|||
|
|||
Slip Sliding Away (Mars Rovers)
In article ,
Keith Lehman wrote: On 10 Mar 2004, Ron wrote: http://marsrovers.jpl.nasa.gov/spotl..._20040309.html Slip Sliding Away Jet Propulsion Laboratory March 09, 2004 On February 9, the wake up song for Opportunity was 'Slip Sliding Away' by Simon and Garfunkel in honor of the Slip Sliding Away was done by Paul Simon after he went solo. So you're saying they never recorded it together? http://www.lyricsfreak.com/s/simon-a...el/124691.html -- Hud Nordin Silicon Valley |
#5
|
|||
|
|||
Slip Sliding Away (Mars Rovers)
In article , mlm wrote:
One wonders how much it would take to enable the rover to figure out that it was not really moving as far as it thought. What about using accelerometer data? Images? Accelerometers that will precisely measure very small accelerations -- the rovers don't move very quickly, and their acceleration is very small indeed -- are complex, heavy, and costly. Not something you put on a low-budget low-mass rover, especially when the need should be rare. And automatic image analysis is really very difficult. Yes, something might be devised for this particular situation... but why bother, unless it's going to happen every other day? -- MOST launched 30 June; science observations running | Henry Spencer since Oct; first surprises seen; papers pending. | |
#6
|
|||
|
|||
Slip Sliding Away (Mars Rovers)
mlm wrote:
Fact is these things are really only elaborate extensions of their engineers egos. Far from the independent robots they are made out to be. Highly capable machines that don't really have a clue. A lot like pro wrestlers. Please show us the robots and decision making software that you have designed, Mr. know-it-all. Yeah. Thought so. |
#7
|
|||
|
|||
Slip Sliding Away (Mars Rovers)
In comp.robotics.misc, mlm wrote:
(Ron) wrote in . com: http://marsrovers.jpl.nasa.gov/spotl..._20040309.html Slip Sliding Away Jet Propulsion Laboratory March 09, 2004 On February 9, the wake up song for Opportunity was 'Slip Sliding Away' by Simon and Garfunkel in honor of the experience the prior day when Opportunity first tried to scoot up the crater slope at Meridiani Planum. The ambitious rover aimed for its target and willingly went forth up the slope; nevertheless, Opportunity slipped and slid in the sand, making it only half way to its target. But ignorance is bliss, and while Opportunity's wheels had a hard time gaining traction on the sandy surface, the rover's brain (or computer) had a hard time grasping that it hadn't successfully made it to its target. The little rover didn't have to encourage itself to make it up the crater slope later by chanting, "I think I can, I think I can," because the little rover thought it actually did make it up the slope the first time. All very nice Ron. Thanks again for sending on this detailed explanation. What is the message to draw from this? The rovers are incredibly stupid. Perhaps, in fairness to the poor things, one should say that they have incredibly stupid decision making software running them. Even the vaunted independent navigation system is incredibly crude when you think about it. One wonders how much it would take to enable the rover to figure out that it was not really moving as far as it thought. What about using accelerometer data? Images? Doesn't the independent nav system look at obstacles in the rover's path? If so, why can't it just figure out that the obstacles haven't changed much despite the odometer clicking off 20 metres (65 feet for the metrically clueless). Fact is these things are really only elaborate extensions of their engineers egos. Far from the independent robots they are made out to be. I would have expected a (or several) more reliable system for measuring distance, such as some sort of radio distance ranging from the base/platform (is it running or does it have any active electronics in it after the rover goes down the ramp?) or Mars sattelite, and/or a downward-facing camera and logic system such as in a modern-day optical mouse, or looking at obstacles as you described. OTOH, there's the engineering idea that the more complex something is, the more likely it is to break down. They probably have opto quadrature detectors in each wheel to detect its direction and speed, which surely also detects stalled wheels and such, so using that for distance measurement is 'good enough.' Highly capable machines that don't really have a clue. A lot like pro wrestlers. Now you've insulted both pro wrestlers AND the rovers! Mark ----- http://mindspring.com/~benbradley |
#8
|
|||
|
|||
Slip Sliding Away (Mars Rovers)
Henry Spencer wrote:
In article , mlm wrote: One wonders how much it would take to enable the rover to figure out that it was not really moving as far as it thought. What about using accelerometer data? Images? Accelerometers that will precisely measure very small accelerations -- the rovers don't move very quickly, and their acceleration is very small indeed -- are complex, heavy, and costly. Not something you put on a low-budget low-mass rover, especially when the need should be rare. And automatic image analysis is really very difficult. Yes, something might be devised for this particular situation... but why bother, unless it's going to happen every other day? Analyzing images in real time still takes a small brain to do. Though insects and small fish can do it, so it can't be as hard as some of the things human brains do routinely. At least the rover didn't tilt or fall over. Can they recover if they turn turtle? Depending on which way, I suppose that instrument arm might be able to help right the rover. Not on its own, NASA would have to tell it how. Hopefully nothing gets broken if it did roll down the slope.... Can the rover, instead of driving straight up the slope, take it sideways? |
#9
|
|||
|
|||
Slip Sliding Away (Mars Rovers)
"Robert Casey" wrote in message ... Analyzing images in real time still takes a small brain to do. Though insects and small fish can do it, so it can't be as hard as some of the things human brains do routinely. At least the rover didn't tilt or fall over. Can they recover if they turn turtle? No. Depending on which way, I suppose that instrument arm might be able to help right the rover. Not anywhere near enough strength where it needs it. And it would ruin the instruments on the end of the arm. If it goes over, that's that. |
#10
|
|||
|
|||
Slip Sliding Away (Mars Rovers)
Greg Crinklaw wrote in
: mlm wrote: Fact is these things are really only elaborate extensions of their engineers egos. Far from the independent robots they are made out to be. Highly capable machines that don't really have a clue. A lot like pro wrestlers. Please show us the robots and decision making software that you have designed, Mr. know-it-all. Yeah. Thought so. I wish I knew it all. I don't Mr. Cynic, but I do know how to solve this problem. And I don't need a multimillion dollar software development budget to do it. Furthermore, the rovers have all the gear they need. Snap photo. Contrast enhance severely to take out gray areas. Move. Snap photo. Measure distance between black spots on photos. Hey presto no movement. Doubt something like this could be done? How about the DIMES system? Yes it was a late addition but it proves the concept. It is a software problem. The hardware to do it is right on the rover. As was mentioned, coming up short on a drive means another day wasted. With a budget for the mission of $400 million, each day is worth a lot of cash. SHould be solvable by a clever engineer with a little extra midnight oil. Don't even need a know it all. Mark |
|
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Space Calendar - November 26, 2003 | Ron Baalke | History | 2 | November 28th 03 09:21 AM |
Space Calendar - November 26, 2003 | Ron Baalke | Astronomy Misc | 1 | November 28th 03 09:21 AM |
Space Calendar - October 24, 2003 | Ron Baalke | History | 0 | October 24th 03 04:38 PM |
Space Calendar - October 24, 2003 | Ron Baalke | Astronomy Misc | 0 | October 24th 03 04:38 PM |
Mars in opposition: One for the record books (Forwarded) | Andrew Yee | Astronomy Misc | 0 | August 3rd 03 04:56 PM |