Mapping the Red Planet with Robot Teams

Sending astronauts to Mars may still be just beyond our grasp, but in the past few decades robots have filled in nicely for us. NASA’s Mars rovers have tirelessly explored the red planet without complaining about the cold or lack of oxygen. And since they do not need food, or much of anything else aside from a bit of solar energy, they have made for a great first wave of explorers as we improve our technology to allow for longer durations of human space travel and return flights from distant worlds.

But we are talking about exploring entire planets, so the few tens of miles that present rovers have been able to cover is still leaving a lot to the imagination. Even after humans arrive on Mars for an extended stay, we will still need more advanced technological solutions to cover the vast stretches of undiscovered land. Just beyond the next hill or under that rock in the distance could be a discovery that changes the course of human history, so there is a great deal of interest in covering as much area as possible in our future explorations.

A high-level overview of the system architecture (📷: S. Swinton et al.)

Researchers at the University of Glasgow believe that the next step in planetary exploration should involve the use of groups of autonomous robots. More robots can without question cover more area, but in any case, the amount of equipment that we can get to another planet is limited. That means that the explorations need to be carefully planned to avoid duplication of efforts, visit the areas that are most likely to be scientifically interesting, and stay away from dangers that could disable the robots and render them useless after those many millions of miles of travel (and many billions of dollars in expenses).

As the number of robots increases, the complexity of the mission also increases. This quickly makes the workload too great for human operators, so higher levels of autonomy are necessary in such scenarios. The team’s unique approach addressed this need through the development of a unique multi-rover mission planner. Beginning with data from HiRISE orbiter imaging, probability distribution maps, and on-board rover sensors, the system creates a plan for exploration of the surface of Mars.

As a first step, the planner leverages data from the Mars Reconnaissance Orbiter to create a map. Using this information, it is then possible to split the area up into regions. Paths that a rover can safely traverse are calculated, then a probability distribution map is overlaid on these paths. This probability distribution map expresses the likelihood that areas of interest will be encountered at various sites along each path — these areas could range from interesting rocks to unexpected anomalies or other features.

A mission plan for a group of robots (📷: S. Swinton et al.)

Finally, an efficient route is calculated for each rover in the group to cover as much area as possible while focusing on the features that are most likely to be of interest. Using this approach, much more of the surface of a planet can be covered when compared with existing methods. This can also be accomplished much more quickly. In a series of experiments, the team found that it was possible to cover an area of over 22,000 square meters in only 40 minutes.

This work also has immediate applications right here on Earth, even before it powers robots on Mars. The researchers suggest that search and rescue operations, for example, could be made more efficient using this approach. But for now, the team is working to refine their algorithms and make them more fault tolerant with the hope that they will one day blast off for the red planet onboard a group of robots.