Sometimes Nature acts as if it is waiting to be explored. For example, it secretly installed "doors" on the icy shell of Enceladus1 – vents through which water vapor and other materials are ejected from the subsurface ocean into outer space. The doors may serve as a natural pathway through the ice crust, which is tens of kilometers thick, to reach the extraterrestrial ocean. The doors may provide us with an opportunity to investigate a unique environment in the vent-conduit systems, which themselves could be habitable. The doors may lead us to the discovery of the second origin of life. The doors may lead us to the answer to mankind's ultimate question – "are we alone?" The doors are just 10 AUs away from us. Why would we leave them open and unexplored? Our concept, Enceladus Vent Explorer (EVE), is a robotic pathfinder mission to enter these doors. EVE's goals are to descend into erupting conduits up to ~2km deep, characterize the unknown interior structure of the vent-conduit system, assess the accessibility to the subsurface ocean through the vent-conduit system, potentially reach the liquid interface, and perform astrobiology and volcanology observations in the vent-conduit system. EVE sends two types of modules: Surface Module (SM) and Descent Module (DM). SM is a lander that stays on the surface, while tens of small (~3 kg, 10 cm in width and 30 cm in length) DMs separate from SM, move to a vent, and descend into it. DMs rely on a power and communication link provided by SM through a cable. As the payload volume of DM is extremely limited, each DM can carry only a single miniaturized instrument. This limitation is complemented by heterogeneity. There are several types of DMs, all of which share the common mobility system but carry different instruments. For example, a "scout DM" creates a 3- D map of the geyser system with its stereo cameras and structured light. A "sample return DM" collects particles and ice cores in the vent and deliver them to the mass spectrometer in the SM. An "in-situ science DM" carries science instruments, such as a microscopic imager and a microfluidics chip for biosignature detection. DMs are sent either sequentially or in parallel. The two greatest difficulties for EVE are the uncertainty in i) the dynamic pressure due to the upward flow and ii) the size of vent. Despite remote observations by the Cassinispacecraft, these two parameters remain poorly characterized. Among the numerous unknowns associated with Enceladus vents, the one that has the greatest impact on the two key parameters is the eruption mechanism. Most hypothetical eruption models that have been proposed thus far fall into either of two categories, which we refer to as the relatively calm "boiling" models, and the more dynamic "cryovolcanic" models. Very roughly speaking, the "boiling" models assume that liquid water boils into vapor under the surface, while the "cryovolcanic" models assume that a fairly pure form of the ocean material ascends the conduit driven by exsolution and expansion of dissolved materials and phase changes. More detailed explanations of the two models will be provided shortly. Existing Cassini observations cannot resolve between the two models. The "boiling" models are more favorable for EVE because they result in a greater vent size and lower dynamic pressure. The worst-case dynamic pressure is ~104 Pa, which is well within the design range of DM. While the vent size is harder to constrain, the "boiling" models give orderof-magnitude greater estimates than the "cryovolcanic" model in general. For example, Kite and Rubin  suggest "slots" of greater than 1-m in width. Therefore, EVE is likely feasible assuming the "boiling" models. The "cryovolcanic" models are more problematic for EVE. Our preliminary analysis suggests that the dynamic pressure and the vent width could be anywhere between 103 - 107 Pa and 1-30 cm, respectively. The worst-case dynamic pressure is beyond the upper limit for DM. While some work suggests 107 Pa dynamic pressure is likely unreasonable for geophysical reasons (D. Hemingway, pers. comm.) and more detailed analysis could better constrain the worst cases, we do not reject these possibilities. Therefore, the feasibility of EVE is undetermined with the current best knowledge under the "cryovolcanic" models. Further study of Cassini data could better constrain parameter estimates of each model. In order to resolve between "boiling" and "cryovolcanic" models, additional orbital observations are likely needed. While observations from orbit around Enceladus could help provide some insight into the appropriate model, uncertainties in interior geometries and vertical profiles of dynamic pressure cannot be significantly reduced without descending into the vent. Therefore, the first mission into an Enceladus vent must face a chicken-and-egg problem: dynamic pressure and vent size remain uncertain until it descends into the vent, but designing a robot to descend into the vent requires knowledge of those parameters. Consequently, in order to maximize the chance of mission success, EVE's DM needs to be designed conservatively in terms of size and resistance to dynamic pressure. Our trade study has determined that the DM configuration that is most robust to the uncertainties in vent parameters is a limbed robot with ice screws as the end effector. An anchoring mechanism is needed because the upward force exerted by the jet is orders of magnitude greater than the downward force exerted by Enceladus's weak gravity (~0.01g). An ice screw is a hollow metal screw used by alpinists for ice climbing. As a byproduct, it produces an ice core when screwed in, which could be brought back by the sample return DMs. Since an ice screw can make a strong anchor with only one wall (as opposed to wedging, for example, which requires two walls), it is robust to uncertainty in the width of conduit. We validated by ice chamber experiments that ice screw end effectors can tolerate the force exerted by ~105 Pa dynamic pressure on the DM. Our trade study also concluded that the most realistic way to power DM is to provide power from SM through a cable because neither RTGs nor primary batteries that fit in the limited volume of DM can provide sufficient energy for the mobility system within a reasonable mission duration. A cable also eliminates difficulties in communication as well as in navigation for the return trip of sample-return DM. A three-section DM can accommodate up to 2 km of cable in its rear section. By adding a dedicated section for storing cable, it can accommodate up to 10 km of cable. The estimated speed of DM is 5.5 m/hr, given 10 W of power for mobility. If continuously operated, it can move 1 km in 7.5 days. We use the proposed Europa Lander as a reference design for SM. The major difference from Europa Lander is the power source. Europa Lander uses a 45-kWh of primary battery, which weighs approximately twice of MMRTG, because the lifetime of the supporting Carrier Relay Orbiter (CRO) is limited to ~30 days due to Europa's harsh radiation environment, thus a longerlived lander mission is difficult to justify (Hand, et al., 2017). In contrast, measurements indicate that the intensity of the trapped radiation environments of Saturn is much lower than the Earth's and is not likely to pose serious problems to spacecraft (Barth, et al., 2003). This justifies the use of RTGs to support a long mission. Given this basic configuration, there is still a vast trade space to explore, particularly in the design of DM – size of the robot, DOF of each limb, choice of actuators, shape, etc. In particular, size is very important. The drag force by the flow is approximately proportional to the crosssectional area of the robot, but the mechanical force that the robot can apply in order to resist the drag also increases with the size. It is beyond the scope our Phase I study to optimize the design of DM. Instead, in order to obtain a reference point for the tolerable dynamic pressure, we developed a prototype design of DM with ~10cm in width. The prototype design only uses existing technologies/materials and commercially available mechanical parts (e.g., motors, gear boxes). It is a point design and not an optimal one, but it has been guided by experienced JPL mechanical engineers. The resulting design has four 5 DOF limbs, actuated by commercially available electrical motors. Each limb features a linear actuator, which is used to move DM's body against the flow. Each limb is designed to support up to 1000 N of force, meaning that DM can withstand up to 4000 N of force with its four limbs. The result of our computational fluid dynamics simulation suggests that the total force exerted on DM with 106 Pa of dynamic pressure is ~7400 N. Therefore, the maximum tolerable dynamic pressure of this particular DM design is 5.4x105 Pa. With safety margins, ~105 Pa would be the safe operational limit for this prototype design of DM. We note that the drag coefficient of the design is ~1.5, while that of typical automobiles, for example, is 0.2-0.3. Clearly there is a significant room for design optimization, particularly in the aerodynamic shape. However, it is beyond the scope of this Phase I study to optimize the aerodynamic design of DM. Beyond the mechanical aspects of the EVE design, our study has considered the nature and degree of autonomy capabilities required to effectively execute the mission. Our study concludes that EVE must be a highly automated system in order to complete the mission within a reasonable duration. One of the most significant challenges for teleoperating DMs is the limited viewshed (geographical area that is visible from a location) in the vent. In Mars rover operations, the drive distance per operation cycle (Sol) of manual planning is limited by viewshed, which is typically ~50 m. In the vent, it would be tens of cm. In case of Europa Lander, each operation cycle is ~24 hr long, which is constrained by the visibility of the CRO. Limiting DM motion to tens of cm per day, it would take decades to move one kilometer. In addition, operating tens of DMs manually would be a major challenge. In order to enable a realistic mission, required autonomy capabilities for DM include 3D mapping, localization, hazard detection and avoidance, route planning, activity planning, self-diagnosis, and recovery from failures. Agile science is highly desired to not miss interesting science opportunities during the traverse. Cooperative autonomy with multiple DMs is also important. The limited computational capabilities onboard the DM are not a limiting factor, since DM can access computational resources on SM through the cable. Furthermore, DMs can communicate with each other via SM. Hence, SM serves as the "information hub" during the automated operation. In summary, the feasibility of EVE depends on the eruption mechanism, which at this point is unknown. Assuming the "boiling" model, in which the expected dynamic pressure is conceptual models for ascent and eruption, and may allow for enhanced interpretation of plume mass spectrometric results. Further exploration of the design space of DM, in particular the optimization of aerodynamic shape and actuation mechanisms, could allow greater resistance to dynamic pressure. Among our set of mid-term goals, resolving between the "boiling" and "cryovolcanic" models is the highest priority. This could be done by a future orbital mission such as ELF (Enceladus Life Finder). Very high resolution imaging combined with extremely sensitive passive emission spectrometry at sub-mm/THz wavelengths at optimized phase angles may give the highest potential for resolving the models. High resolution imaging of individual vents will also help identify the landing target for EVE. Finally, we would like the readers to recall the excitement of Jules Verne's Journey to the Center of the Earth. In the classic science fiction novel, Professor Otto Lidenbrock and his company descend into an Icelandic volcano, discover a vast subsurface ocean, and encounter exotic life. In the near future, EVE could make such a fictional adventure real, not on Earth but in the frosty and mysterious world of Enceladus.