Title: Autonomous Mars Rover for Geological Sample Collection Crew Member:Vindhya Ganti, Crew 326 Green Hab Officer Objectives: Train an image-based navigation system on local landmarks to allow a rover to navigate autonomously Description: To make the rover autonomous, first the rover needs to be able to identify different locations and have some sort of mapping system to identify where it is relative to other locations. To accomplish this task, a teachable machine learning image algorithm was fed roughly 20–35 images of three different locations (Hab, White Rock Canyon, and Kissing Camel) from various angles. Furthermore, images that didn’t fit any of those three categories (Unlabeled) were mixed into the dataset. Then images were preprocessed with a noise filter to build tolerance to unideal conditions like low lighting and dust. Then, annotations were assigned per image, identifying the different types into classes: Kissing Camel, White Rock Canyon, Hab, Unlabeled. In each class, images were assigned into three different types: train, valid, and test, to teach the model, give feedback to the model for reiteration, and completely new to the model to test it.
Preliminary Results and Analysis: The model differentiates between the Kissing Camel region (both east and west ridges) and the HAB with 92.76% accuracy, calculated from the test type. Additionally, the model differentiates between random photos of other regions and both of the locations.
Recommendations/Next Steps: It would be ideal to connect imaging logic with the hardware of the rover. This means using the correct label and distance from the HAB, and then having the rover have a camera setup to autonomously move in the direction of the HAB.
Title: Autonomous Mars Rover for Geological Sample Collection Crew Member: Daria Bardus, Crew 326 Journalist Objectives: The preliminary testing involved creating a rover that could be maneuvered with a joystick controller to collect a sample for later measurement and observation using a basic static scoop mechanism. Description: In total, three soil samples were collected from different areas around the Mars Desert Research Station (MDRS), including Kissing Camel Ridge, White Rock Canyon, and the area immediately outside the HAB. At each location a different soil type was identified to test the rover’s ability to collect varied samples. A two-foot collection zone was measured and marked in front of the rover. The driver station was then used to start the program, and the joystick controller was used to first lower the scoop on the linear actuator and then drive the rover forward. When the rover had traveled approximately two feet, the scoop was raised and the sample was then placed into a bag for later measurement and observation. Preliminary Results and Analysis: On average the rover was able to collect a sample of 47.78 g, which is approximately 42.56% of the scoop, this was considered a success. During testing, hard, compact soils were difficult to collect due to the scoop not being able to break through the surface without getting stuck. Conversely, the wheels would dig into the soil if it was too soft. This could be fixed by adjusting the power of the motors in the drive train, but this could not be done while on EVA. Recommendations/Next Steps: In the current iteration of the rover, the scoop is made of 3D printed PLA+, but changing the material to something stronger would help with collection and would also help the rover overcome obstacles in the terrain. Increasing the size of the scoop would allow the rover to collect more soil. A collection system that is not static and could store the sample in the rover would improve the feasibility of this project, ideally a conveyor belt system would be used.
Title: Remote sensing for ISRU Crew Member: Cesare Guarniniello, Crew 325 Commander Objectives: Utilize remote sensing tools for the evaluation of geotechnical properties of the terrain for potential In-Situ Resource Utilization (ISRU) for construction. Description: Using spectroscopy, geological regions of interest will be examined for their properties that would determine their use for construction. Properties of interest are water content and bulk size of particulates (ie sandy vs rocky).
Samples of clay rock were collected from Compass Rock, Somerville Outlook, Barrainca Butte. Samples of shale rock were collected from Skyline Ridge. Preliminary Results and Analysis: Samples are currently being processed at back at Purdue.
Recommendations/Next Steps: Collecting samples from more diverse geological regions around MDRS. Determining different analysis techniques to identify more properties of regolith before collection. Determining ways to test collected regolith at MDRS station.
Title: ISRU Materials for Construction Crew Member: Banjamin Huber, Crew 326 Scientist Objectives: Gather materials from the surface of Mars to make bricks for construction and testing the strength of those bricks Description: Three small starch based concrete samples out of regolith from the MDRS region were created from different geologic locations. One brick compound uses potato starch as a binder and the other compound uses potation starch and Jerusalem Artichoke inulin mixture (6 g of starch, 2g of inulin), 2 bricks of each compound were made. These bricks were then strength tested using a rebound hammer. The general method of creating the bricks involves homogenizing the geologic samples and mixing them with the respective compounds, activating the starch in the lab oven, a series of cooling and drying phases, and then compressing and dehydrating the bricks before testing with a rebound hammer.
The sample collection location were chosen based on geologist reports of the region. The first was a dark red clay, indicating the presence of hematite, from along Hab Ridge Road. The next sample was a lighter red clay taken from Phobos Peak. The third was a sandy streambed material from halfway into Candor Chasma. Preliminary Results and Analysis: With the sample collected from Hab Ridge Road, the inulin and starch mix had small starch granules that clumped together and hardened. After the dehydration and drying process, the bricks were significantly cracked and still moist in the center, which resulted in the bricks being able to only withstand three tests until failure. During homogenization the clay did not mix well with the water and became sticky. After the bricks were formed and dehydrated, the outside had small cracking and salt had settled on the surface of the bricks.
The Phobos Peak bricks stuck to the surface of the mold, which resulted in weakening prior to the final dehydration step. After multiple tests, the bricks were more cracked and significantly weakened and ultimately failed after 3 strength tests.
The sample from Candor Chasma had no cracking, but the rebound hammer made large indentations at the test site. These bricks were able to survive more than 3 strengths tests.
The sand brick is seen to be the preferable choice as it has the highest strength values for both compounds as well as no cracking when drying. Recommendations/Next Steps: There were some difficulties with this research while at MDRS. The whole brick making process had to be adjusted throughout the mission, to account for equipment differences and power limitations. It may be of some value to repeat strengths test on the earlier bricks with an updated process. There are still many different geological sites that to collect samples from. A mixture of samples, such as 60% sand and 40% clay, would be of interest as well.