Rocky 7 - Mojave Desert Field Experiments May 1997 Science Plan, Results, and Lessons Learned

Rocky 7 underwent its second major field test at the end of May, 1997. Setup and engineering tests took place from May 22 through 24, and field testing with the scientists occurred from May 25 through 29. On May 30, members of the science and engineering teams visited the March 1998 test site at Silver Lake while students remotely controlled the rover at Lavic Lake.

For the May 1997 tests, the rover was deployed to the Sunshine Flow, Lavic Lake, and an alluvial fan. The rover started on the flow, the surface of which was covered with a desert pavement of basalt cobble underlain by approximately 1 m of aeolian silt, and traveled down onto the cratered surface of the playa. The rover finished its traverses on an alluvial fan (shown in the figure below). The May 1997 field tests allowed Rocky 7 to perform over a wider variety of terrain with more Mars-like characteristics than in the December 1996 tests. Rocky 7 traversed just over 1 km during the May tests and rehearsals that occurred in April 1997.

1. Traverse 1 to 2 km using waypoints and science targets selected from descent and panoramic imaging. Acquire measurements at a number of science targets. Extend the ground operations to remote sites at Ames and JPL to help define traverses and targets.

2. Deploy a Mössbauer (MB) spectrometer and a nuclear magnetic resonance (NMR) spectrometer in a small box on the rover's mast, and a Point Reflectance (PR) Spectrometer. The PR spectrometer is located on the rover's manipulator arm. A variety of science targets (rocks) were to be examined with the spectrometers. These included basalt samples with different types of coatings (e.g., dark gray and red desert varnishes), a felsic sample from the alluvial fan, and a sample from the playa. Since the individual MB measurements require several hours per sample to complete, they were to be taken overnight. MB and NMR data were to be collected separately by swapping out the instruments, i.e. only the MB or NMR was in the mast at any given time.

3. Use the stereo imaging systems to explore sites of interest along the traverse route. This was to include obtaining images of the pavement and accretion mantle on the flow, the playa, and the craters. Also of interest were drainage features present on the alluvial fan.

Rocky 7 was commanded by an operations team consisting of scientists and engineers on site and at remote locations at JPL and NASA/Ames. The on-site operations team was sequestered in a trailer, unable to see the rover in action, using only rover-provided data and "descent" images from the helicopter flights, in order to simulate conditions during rover operations on Mars. Several other teams accompanied the rover on its traverses: a SPICE team took total station measurements of the rover's location to determine how accurately it knows its position; instrument teams for the NMR and MB spectrometers placed their instruments on the rover's mast and took measurements of science targets; and the Rocky 7 development team was on hand to observe and to provide technical support. Other personnel took notes and photographs, videotaped the rover, and collected samples examined by the rover for further tests and documentation.

Results

Although the actual traverses differed somewhat from those planned using the March 1997 descent images, all major objectives were achieved. Specifically, the following tests were accomplished:

1. Conducted rover testing on three different terrains: a lava flow, a cratered playa, and an alluvial fan. Total rover traverse was 1058 meters, including April rehearsals. A high volume of telemetry data was acquired, together with total station surveying observations, to generate SPICE files that delineate locations and orientation files for the rover as a function of time.

2. Conducted four science experiments: (1) Sunshine Flow varnished basalt rock experiment, (2) Sunshine Flow aeolian accretion mantle experiment, (3) Mud-cracked crater floor experiment, and (4) Alluvial fan surface experiment. The experiments included such operations as science imaging, placing a mast-mounted science instrument, PR spectrometer pointing and data acquisition, and soil sampling. Mast-mounted instruments were NMR and MB spectrometers. One other instrument mockup, an X-ray diffractometer from Ames, developed by John Marshall, was mounted on the rover for configuration evaluation.

3. Rocky 7 rover performed well in the rough natural terrain, dusty and hot environment. Temperature at the desert reached well above 100 F at times.

4. Performed remote operations from JPL (with Jeff Plescia and Dan McCleese) and NASA/Ames (with Jack Farmer). This operation was only partially successful. Jack Farmer's computer was not 100% compatible with WITS and not all requested operations were possible through WITS.

5. Six schools, including one from Finland, participated to cooperatively control Rocky 7 remotely. Students actually commanded the rover. This operation was very successful and received considerable press coverage.

Some statistics: Rover Traverse: 1058 m Science experiments: 4 Images returned by the rover: 519 Rover mast deployments: 89 Instrument placements: 9 Equivalent Mars days operations: 32

Overall the experiments went well, considering the extent of the traverses, the number of deployments, and the amount of data returned by the rover. The following lessons highlight problem areas on which the science teams and rover developers will focus in preparation for the spring 1998 field experiment.

1. Basalt pavement, playa, cratered playa, a crater proper, and an alluvial fan were traversed. The site for next spring at Silver Lake contains rock fields as dense as those observed at VL2, which will present hazard problems for Rocky 7. The lesson is that we have demonstrated proof-of-concept on hazard avoidance using stereo models, but we still have a long way to go to be able to traverse rocky terrain on Mars. Further work on sensor processing, reactive collision avoidance, and path planning is needed.

2. At one point Rocky 7 encountered a major difficulty when it became stuck on a sharp basalt cobble. The operators commanded Rocky 7 to move in reverse, but it remained stuck. Finally the rover had to be manually lifted from the cobble because of the danger that it might become damaged. The operators believe they could have freed Rocky 7 by commanding it to move sideways, if there had been more time to analyze the situation (as there would be during actual Mars rover operations). This event revealed a logic flaw in the command software. Rocky 7 was able to identify the sharp cobble as an obstacle to avoid, but the cobble had also been designated by the operators as a waypoint. The obstacle avoidance system failed to override the rover's attempt to approach the waypoint.

3. The self-inspection images that were acquired when the rover got stuck on the rock were interesting but did not include the underbelly. It was also difficult to understand which images corresponded to which areas of the rover. The self-inspection option needs to have a capability of seeing the underside of the rover. Include in WITS a self-inspection window that shows a rover icon and small views of each self-inspection image. Double clicking on the images should enlarge them for better viewing.

4. The May experiments were time-consuming and costly. The next field deployment should be done with fewer people and less infrastructure in the field. More should be done remotely. For the March 98 experiments, use a blind test mode in which a distributed team unfamiliar with the Silver Lake site has simulated orbital and descent images and areas marked as out of bounds for the rover (because of bushes, etc.). They would be given the objective of testing the hypothesis that Silver Lake was once an actual lake, using imaging, point spectroscopy, and in-situ observations. Have just a few people in the field and let the rover operate for a longer period of time, perhaps several weeks. It is important at the end of the tests to have key people involved in planning and analysis of the sequences walk the traverses as part of the lessons learned. In this way one gets an appreciation for the differences between what a person can see and what the rover sees.

5. WITS needs more development and testing before the next field deployment, as does the ability to conduct remote planning through WITS. Specifically:

   • The first thing scientists will probably want to do after landing is a 360-degree color pan to see where they are and where they want to go. Color pans could not be used easily in WITS, so we started off on the wrong foot.

   • The pan and nav images are acquired in stereo and should be viewable as a WITS option in stereo to help in planning and analysis. Incorporate anaglyph or polarized viewing capabilities for planners.

   • Non-rectified descent images were used in WITS due to a lack of time to integrate the rectified images. WITS should be able to accept the rectified descent data, provide distances and bearing interactively, and display simulated pan images generated from the descent images.

   • Sequences for commanding the rover should be implementable in modular forms, such as macros, in which time and location are updated. Use the results from the May tests to help update the utility of WITS.

   • Remote operation with WITS by the schools went well, but remote operation by Jack Farmer at Ames Research Center did not go well. A number of reasons contributed: lack of training on WITS, incompatible systems, the satellite link to the desert came and went, and things were done that WITS could not easily support. Make sure that users are trained on WITS and that they have tested WITS on their computer configurations before the field test.

6. A physical properties experiment was done by locking all but one wheel and spinning the operating wheel to dig a hole. During the March 1998 tests, record motor currents during these operations to be able to monitor shear stress required to dig. This measurement is crucial for understanding soil strength.

7. The focal distance for the pan cameras needs to be commandable and/or the spatial resolution of the cameras needs improvement. Features tended to be blurred if they were more than about 5 meters from the rover.

8. The surface sampler arm and reflectance spectrometer need to be commandable by the operator to any location within the field of reach of the arm, not just to 4 positions.

9. For the March 1998 tests the rover should have a reflectance or emission point spectrometer, boresighted with the pan imaging system, that can identify key minerals in addition to determining color and morphology. For these tests we should also have an instrument (e.g., Marshall's XRD or a Raman spectrometer) on the mast that can make actual real time, in-situ measurements of mineralogy for samples as an integral part of the experiments.

10. Although soil samples were collected we have yet to address the difficult problem of collecting and storing rocks. This should be attempted in some fashion during the spring 1998 tests.

11. One of the key lessons learned is that Mars rover objectives have to be well matched to the data-limited environments in which rovers will operate. For example, we were able to determine from the rover data that we were on a basalt flow covered with a desert pavement, but we missed some missiles that were present. Why? The flow and pavement are really extensive and the rover, despite limited data, was able to observe the properties of these features and materials because they were widespread. The missiles were rare occurrences and thus they are unlikely to be observed in the rover data-limited environment. The point is the importance of site selection to ensure high probabilities of finding the targets of interest and the importance of descent imaging for planning traverses to regions which have an abundance of science targets.