Rover Autonomy Study #2

Which autonomy software technologies would provide the best ROI for missions to explore mid-latitude and polar regions of Mars with rovers?

This pilot study for the Computing, Information, and Communications Technology Program (CICT) was based on two potential missions to Mars -- an MSL-type mission to Isidis Rim near the Martian equator, and an AFL-type mission to the north polar region. MSL is the Mars Science Laboratory mission planned for around 2009. AFL is the Astrobiology Field Laboratory, an advanced mission concept for around 2013 to 2019. The primary scientific objective of each mission is to determine the capacity of the environment to sustain life.

The study's objective was to develop and demonstrate a system for evaluating and comparing groups of advanced autonomy software technologies, most of which are currently at the R&D stage (TRL 2 or 3), in terms of the impact they would have on science return in each of the two missions.

We began with a top-down methodology in which we derived, from top-level mission goals, the technology capabilities that would be required to operate at the two disparate sites and produce the desired science. Then we went bottom-up, taking a group of technologies, determining their capabilities, and evaluating the potential impact that enhanced autonomy would enable them to make.

Deriving capability requirements

Following the top-down methodology, we decomposed the mission goals into their constituent functional requirements. For example, the science requirements call for such capabilities as mobility, instrument placement, sample acquisition, and telecommunications. Mobility, to take one of these capabilities, entails range-mapping to estimate distance to nearby objects and possible hazards, path planning, and obstacle avoidance. Obstacle avoidance, in turn, breaks down into obstacle detection and navigation. Using this system, we ultimately arrived at a comprehensive list of functional requirements that could benefit from enhanced autonomy provided by the technologies under consideration.

Autonomy Technologies Selected for Evaluation

Turning to the "bottom-up" portion of our methodology, we selected 15 technology groups as representatives of the diverse interests of the CICT program and of the seven technology areas that enable surface operations.

Area	Technology Group
1 and 2	Onboard Fault Identification for Planetary Rovers
1 and 2	System-Level Verification Technology
1 and 2	Autonomy Infusion Simulation Environment
3	Distributed Control Testbed for Autonomy
3	Rover Autonomy Architecture
4	Single-Cycle Instrument Placement
4	Rover-Based Manipulation
5	Multimedia Human Computer Interfaces
5	Human-Centered Computing for MER
6	Onboard Science Analysis
7	MER Rover Sequence Generation
7	Contingency Planning for Concurrent Activities
7	Accelerated Long-Range Traverse
7	System for Mobility and Access to Rough Terrain
7	Super-Resolved 3-D Surface Models from Rover Images

Key to Areas:

Fault Management Validation/Verification Software Architecture Approach/Instrument Placement

Human-Computer Interaction Sample Handling Mobility

The task before us was to match their capabilities with the capability requirements derived from the mission goals, and to determine the extent to which each of these technology groups would produce greater science results if it had enhanced autonomy.

Calculating Impact on Science Return

Through interviews with experts, we developed performance parameters for each of the technology groups, and determined which elements of the mission the technology would help.

For example, we determined the impact that "System for Mobility and Access to Rough Terrain" would have on the rover's traverse rate. Then we plugged that information into the Mission Model to calculate its impact on the number of sols (Martian days) this technology would save over the current state-of-the-art, as a percentage of the total mission duration.

This template illustrates the procedure for calculating how many sols each technology group would save, as a percentage of the total mission. Each technology would impact (and presumably improve) one or more of the bubbles that lead to determining a number for "sols." "Nominal sols" refers to the number of sols that would be spent using state-of-the-art technology as represented by the MER rovers.

Saving sols means enabling the rover to spend time doing science instead of -- in this case -- traveling. So saved sols are presumed to correlate to an increase in science value. We used the number of saved sols in the ROI Model to determine each technology's return on investment, and the resulting numbers were used to rank the technologies.

This template illustrates the procedure for determining ROI for each technology group. "Increase in science value" uses the "percent sols saved" number calculated in the Mission Model.

The table below shows the results of the initial prioritization. The task names have been replaced by the letters A-O because the data is still preliminary and under review. We provide the results table, with the data and supporting models, to all parties involved to begin a dialogue on the perceived impact and rationale.

Combined ROI used illustrative weighting with relative ratio 2:1 for MSL and Polar missions.

ROI represents increase in science value (as measured by the number of sols saved over SOA) divided by cost. When calculating the combined ROI for each technology task, we gave the MSL value twice as much weight as the polar value. This weighting is somewhat arbitrary and could be changed if desired. But it was intended to reflect the fact that these technologies are more likely to be used in the more-imminent MSL mission, and to be precursors to the technologies that will enable and enhance the polar mission. Though these technologies are innovative, far exceed SOA for the most part, and are intended for long-term impact, they will have as much as a decade for further improvement between the two missions.

Note also that these ROI numbers are not intended to represent final, definitive evaluations, but rather a solid basis for further investigation and discussion. They indicate the potential performance of each technology under certain conditions and for specific purposes. A given technology might benefit additional operations that, if factored into the study, would improve the technology's ROI. Similarly, we could amplify the study by factoring in additional metrics -- such as development and operations cost, heritage value, innovation, and public inspiration -- and potentially arrive at different results.

However, the study does demonstrate that it is possible to estimate mission-level science return impacts of diverse autonomy technologies, that the results can be very useful in assisting decision-makers in the selection of technology groups for funding and development, and that these methods are applicable to a wider class of technologies and mission classes.