Robust Autonomy for Planetary Sampling

Surface exploration of Ocean Worlds presents novel and unexpected challenges for robotic landers. Several aspects of the proposed missions to worlds such as Europa, Enceladus, and Titan create a need for a higher degree of autonomy than that required by previous uncrewed exploration spacecraft. These include the greater physical separation from Human operators, the more dynamic aspects of the environments, and tighter power budgets of the platforms. These challenges not only complicate the execution of tasks in the environment, but also may affect the exploration platforms in unexpected ways.

Honeybee Robotics (Honeybee) will (1) develop a robust autonomous task control framework which extends NASA’s PLEXIL (Plan Execution Interchange Language) executive with stochastic decision-making capabilities, (2) implement alternative optimal and sub-optimal manipulator sampling policies, and (3) integrate and demonstrate the novel autonomy framework with the simulated OceanWATERS testbed for NASA’s Autonomous Robotics Research for Ocean Worlds (ARROW) program. The end result will both advance the robustness of ARROW autonomous task-level control and provide new capabilities to a core NASA software system.

The proposed Stochastic-PLEXIL (SPLEXIL) architecture will expand the PLEXIL language and executive to maintain and estimate probabilities of success of PLEXIL nodes based on their children and input parameters. The proposed approach is based on recent results from the self-driving vehicle community, whose members have focused on developing lightweight and predictable task-level controllers which can handle uncertainty. Specifically, Multi-Policy Decision-Making (MPDM) is an approach which frames the task-level control problem as a Partially Observable Markov Decision Process (POMDP) policy optimization problem. POMDPs are frequently used to formulate autonomous controllers with a high level of operator independence, and the use for robotic systems such as uncrewed landers has been studied for decades. The innovation of MPDM is its policy selection algorithm which approximates a POMDP with minimal computational overhead and tractable development design patterns.

Two alternative sampling manipulation policies will be developed within the SPLEXIL framework, in order to demonstrate automatic and proactive policy selection in the OceanWATERS system under degraded operating conditions. The primary belief states which will affect policy selection will be environment confidence and actuator control performance. The baseline plan is to develop two mid-level sampling behaviors (PLEXIL actions). These two behaviors will achieve the same goals of moving the sampling tool to a point in the environment, extracting material, and moving it back to an analysis tool. One will rely on vision and precis position control, and the other will rely on actuator mechanical feedback.

This effectively transforms a given PLEXIL plan into a hierarchical policy which can optimally select which task nodes to execute based varying operational conditions. This approach will both leverage the heritage capabilities of the PLEXIL executive and plan development workflow while increasing the robustness of the task control subsystem to unexpected faults and degradations. This practical approach to robust autonomy will increase the efficiency of unsupervised surface operations, which is necessary for time- and resource-constrained missions to these unexplored Ocean Worlds.

Developing Instrument or spacecraft technology to improve measurements for future planetary science missions.

This is early/mid-stage (TRL 4) — the most realistic path in is NASA SBIR/STTR, which funds small businesses and research institutions to develop technology aligned with NASA's needs (equity-free, phased funding). Check whether a current SBIR/STTR solicitation topic overlaps with this project's technology area, or contact the project directly (above) to ask.

None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.