We developed a scientific exploration subsystem involving a three DoF secondary manipulator and analysis chamber to categorize any soil and rock sample into extant, extinct or ‘ No Presence of Life’. Mesh of multiple end effectors and a circular scoop motion can effectively collect up to four soil samples preventing any cross-contamination alongside the uv lights in the funnel. The samples are each deposited into 3 isolated beakers for the detection of biomolecules with chemical tests, which involve: Protein using ninhydrin solution, Lipid using Sudan III solution and Carbohydrate using benedict's solution. Qualitative color changes are observed through a camera and the results are then streamed back to the base station for immediate evaluation and feedback as input to our customized algorithm, classifying the sample into Extant, Extinct, and/or No Presence of Life. USB microscope feedback along with the presence of mVOC sensors have been used to detect whether a rock sample is Extant. We also developed a pipeline of multiple binary classifiers trained on a custom dataset of 765 rock samples involving deep neural networks and transfer learning to effectively classify rock samples without any physical contact. The outcomes of the sensor feedback and model predictions have been used to classify a rock sample. A full mock test has been conducted to evaluate the performance of the sample acquisition and analysis process.