Rover Navigation

Moving around Mars

The rovers were designed to trek up to 100 meters (about 110 yards or 328 feet) across the martian surface each martian day, though they have gone much farther. While a complete martian day (called a sol) is about 24 hours and 40 minutes long (or 24 hours 37.5 minutes if you prefer), the Sun can only provide enough power for driving during a four-hour window around high noon. That means the rovers have to be able to move quickly and effectively.

Moving safely from rock to rock or location to location is a major challenge because of the communication time delay between Earth and Mars, which is about 20 minutes on average. Unlike a remote controlled car, the drivers of rovers on Mars cannot instantly see what is happening to a rover at any given moment and they cannot send quick commands to prevent the rover from running into a rock or falling off of a cliff.

During surface operations on Mars, each rover receives a new set of instructions at the beginning of each sol. Sent from the scientists and engineers on Earth, the command sequence tells the rover what targets to go to and what science experiments to perform on Mars. The rover is expected to move over a given distance, precisely position itself with respect to a target, and deploy its instruments to take close-up pictures and analyze the minerals or elements of rocks and soil.

Engineers and the rover itself, however, have to accomplish quite a few things:

Understand Distance

The rover has a difficult time knowing exactly how far it has traveled, where it has been and where it is.

For example, if the flight team asks the rover to move forward 100 centimeters, turn right, then extend its robotic arm and analyze a rock, the rover will follow the commands in reference to its current location. What would happen if the rover couldn't see, and had to rely just on its wheels to tell where it had moved?

Like a car on Earth, the rover uses its odometer to click off the distance it has traveled. If one revolution of the rover wheel equals 25 cm, then after the wheels have revolved four times, the rover should technically have moved forward 100 centimeters (25 cm X 4 = 100 cm). But, unlike cars on Earth, the rover doesn't drive on smooth, paved roads. The rover moves on rocky and sandy martian terrain.

The rover wheels might have a hard time grasping onto the loose-gravel ground. The wheels could spin in place before they actually gain tracking. So if the wheels spin four times before they find firm footing, the odometer will read 100 centimeters, and the rover will stop. Thus, the rover will believe it has moved forward 100 centimeters, when in reality, it hasn't moved at all and may have dug itself into a rut instead. Without other safety checks it might then turn and bang its wide solar panel wing into a rock behind it (that wouldn't have been in the area if the rover had moved forward). The rover would then continue to follow the chain of commands and extend its robotic arm, hoping to meet the rock 100 centimeters from where the rover began its "trek." The lonely robotic arm, alas, would be flailing in the wind, never finding its appointed rock - or worse: without camera-enabled safety checks the arm could again firmly crash into another rock it should have safely passed by along its 100-centimeter journey.

Imagine yourself being given a command to walk from your bedroom to your kitchen, and the only way to get there was to follow these rules:

  1. Scan the area and the route in front of you.
  2. Close your eyes and wait at least 20 minutes.
  3. Keeping your eyes closed, walk to the kitchen without hitting anything.

Now imagine how much easier it would be to get from your bedroom to your kitchen if you could open your eyes every 30 centimeters (1 foot) to reassess the situation.

Avoid Hazards

To safeguard against the rover getting lost or inadvertently crashing into unexpected obstacles, engineers have developed software to help the rover make its own safety choices and to "think on its own." The rover hazard avoidance software stops the rover on an average of every 10 seconds, reassessing the situation and computing its next move for the next 40-50 seconds, after which it begins traveling again. With the hazard avoidance software, the rover can travel safely an average of 30 centimeters (1 foot).

How much does it see? The rover uses pairs of Hazcam images to map out the shape of the terrain as far as 3 meters (10 feet) in front of it, in a "wedge" shape that is over 4 meters wide at the farthest distance. It needs to see far to either side because unlike human eyes, the Hazcam cameras cannot move independently; they're mounted directly to the rover body.

Who chooses where it should go? Using visuals previously downlinked from the Navcams, the scientists and engineers give the rover a "go to waypoint" command, which includes the distance and heading of the intended destination. Later commands give the locations of the targets where the rover should deploy the science instruments. So people tell it what its goal is, but the rover chooses the best way to get there.

The "go to waypoint" command also contains a coordinate "boundary" of acceptable distance from the target. This boundary gives the flight team a high level of confidence that the rover will avoid driving on or over the target or into dangerous terrain that was out of the line of sight from the Navcams, and also prevents the rover from spending too much time getting to an "exact" location, since it might have slipped a little anyway.

The "go to waypoint" command includes time and distance limits that protect both against missing the target and also against being diverted too far from the goal by obstacles. If the ground is known to be safe, something like15 to 25% of the line of sight distance (relative to the target) could be given as the limit. For example, if the target is 60 centimeters (2 feet) away on flat surface, and the rover is still moving after 2 minutes (1 minute longer than the trip should have taken), something most likely is wrong and the rover will stop. Fixed time limits are also given to the rover wheels for each "step" that the rover takes when going to a waypoint. If the rover were to get stuck on a single step, this limit would prevent it from running the wheel motors until the overall timeout for the "go to waypoint" command expired (which could potentially be very long, perhaps tens of minutes to hours for far field traverses).

The autonomous hazard avoidance software works in conjunction with the Hazcams, wheel odometry, and the Inertial Measurement Unit (IMU).

Create Maps

While the rover is stopped, a pair of images from the front (and occasionally also the rear) Hazcams is captured and processed, into a set of (x,y,z) coordinate points in front of the rover. This resulting map of points is processed into a set of terrain features (steps, slopes, roughness) which serve as a three-dimensional model of the actual terrain in front of the vehicle. The Sojourner Rover on the Pathfinder mission in 1997 took 20 separate measurements for every step. The Mars Exploration Rovers take between 6,000 to 10,000 points of measurements per step.

This model is used to determine if the terrain features represent obstacles for the rover (for instance, a feature with height of 30 centimeters (12 inches) or greater is considered an obstacle). A small number of short potential paths in the direction to the destination are checked within this model and a safe path avoiding obstacles is chosen.

The rover moves a short distance (about 30 cm/1 foot) along this path and the process is repeated. As new terrain models are acquired, they are organized into a "world" model with the rover at the center of approximately a 10m-by-10m area. Due to the possibility of wheel slippage, the rover is programmed to only remember a small map around it in order to remain accurate about the surrounding obstacles. Information about the terrain surrounding the rover beyond the 10m-by-10m area therefore must be deleted every additional 5 meters driven. So, as the rover goes forward (roughly six times the length of its own body), it forgets where it has been.

Some of this map data is stored elsewhere onboard the rover and some is sent back to Earth to help build a master map; however, there is a limit to how much data can be sent back to Earth. The amount that the rover can store is more than the amount it can uplink to Earth so the flight team must choose which data is sent ­ science or rover-related. (For further information about data flow to and from the rover, please see Rover Communications.)

Once constructed, the map helps the rover select safe movements to the destination and prevents the rover from encountering obstacles already avoided during prior segments of the drive. The rover proceeds to move to the destination until the accepted distance near the destination is achieved, or (in failure cases) either the rover has driven a distance farther than the target or for a time which exceeds that accepted for the path.

Keep Balanced

The rover's Inertial Measurement Unit (IMU) uses gyroscopes and accelerometers to determine the heading and tilt of the rover. The gyroscopes measure small heading changes very accurately, and the accelerometers measure where gravity is strongest, pulling down on the rover. Having knowledge of where gravity (down) is, the rover can partly assess its orientation. The rover also uses its tilt sensors to prevent rollover.

If the IMU fails or the rover is lost, the flight team can use the Pancam, which is not part of the autonomous system, to try to figure out the rover direction and position relative to the horizon (attitude).

Know Direction

To determine the rover pointing direction, the Pancam sweeps the sky until it finds the sun. Then, by patiently staring at the sun for 10 minutes or so, the sun tracks approximately 2.5 degrees across the sky.

The flight team can then compute which way the rover is facing by using the sun's movement and correlating it to the known time and date. For example, on Earth in the summer at noon, the sun will be high in the middle of the sky. Depending on the direction the sun moved in 10 minutes, you could tell which direction you were facing since the sun moves across the sky from east to west.

Traverse Far and Well

The maximum achievable traverse distance depends on how early the traverse is begun, the amount of energy available for a traverse (the solar panel power degrades over the life of the mission as Mars gets farther away from the sun or as dust accumulates), the terrain type (rocky or smooth), and the time limits given by the flight team or the timeouts automatically incurred along the traverse.

Although the rover is capable of traveling up to 100 meters (328 feet) per sol, a "safe" traverse (where humans can see the rover's path ahead of time) in terrain equivalent to where Viking 1 landed most likely will be around 40 meters (130 feet) in a single sol.

To learn more about how the scientists and engineers select where the rovers will go, how they will get there, and what the rovers will do each martian sol, please see: Science Operations.

To learn more about how the scientists and engineers select where the rovers will go, how they will get there, and what the rovers will do each martian sol, please see: Science Operations.