Artemis-Q

Inspiration

ARTEMIS-Q was inspired by the complexity of crewed space missions, particularly NASA’s Artemis II mission. Unlike uncrewed missions, crewed trajectories must balance not only fuel efficiency and timing, but also human safety under uncertainty, including radiation exposure, communication constraints, and system failures.

We were motivated by a broader question:

How can we design a system that does not just optimize a trajectory, but determines whether a mission should proceed at all?

What It Does

ARTEMIS-Q is a crew-aware mission planning and decision support platform that evaluates spaceflight trajectories under uncertainty. It integrates optimization, simulation, and decision logic into a single system that considers both technical and human constraints.

The platform models:

Trajectory feasibility under resource and communication constraints Radiation exposure and crew health risk Mission uncertainty through stochastic simulation Operational decisions such as continue, replan, or abort

Instead of optimizing for efficiency alone, the system evaluates whether a mission is safe, feasible, and justifiable for human spaceflight.

Quantum-Inspired Optimization

Trajectory planning is formulated as a Quadratic Unconstrained Binary Optimization (QUBO) problem. Each possible trajectory is encoded as a binary decision space, allowing the system to evaluate complex tradeoffs across multiple objectives.

We minimize a weighted objective:

$$ J = \lambda_f \cdot \text{Fuel}(x) + \lambda_r \cdot \text{Radiation}(x) + \lambda_c \cdot \text{Communication Penalty}(x) + \lambda_s \cdot \text{Safety}(x) $$

This formulation enables the use of quantum-inspired optimization techniques, including annealing-based search heuristics. These methods allow efficient exploration of large combinatorial solution spaces and help avoid local minima when evaluating competing mission constraints.

While implemented classically, the approach is compatible with quantum annealing paradigms and reflects how quantum systems approach optimization problems through energy minimization.

Uncertainty Modeling

To account for real-world unpredictability, ARTEMIS-Q integrates Monte Carlo simulation directly into the evaluation pipeline. The system simulates:

Stochastic radiation events
Communication disruptions
Variability in mission conditions

Each trajectory is evaluated across many sampled scenarios, producing distribution-aware outcomes rather than single deterministic results.

Crew Health Risk Modeling

We introduce a Crew Radiation Readiness Index (CRRI) to quantify mission viability from a human safety perspective.

Total radiation exposure is computed as:

$$D_{\text{total}} = \sum R(i, t) \cdot \Delta t \cdot S$$

Peak exposure is defined as:

$$D_{\text{peak}} = \text{max } R(i, t)$$

These are combined into a risk score:

$$\text{Risk} = \alpha \cdot D_{\text{total}} + \beta \cdot D_{\text{peak}} + \gamma ·T_{\text{unsafe}}$$

This score is integrated into the optimization objective, ensuring that human health is treated as a core constraint rather than an afterthought.

Decision Intelligence

Beyond trajectory evaluation, ARTEMIS-Q supports mission-level decision making. For each scenario, the system evaluates:

Continue
Replan
Abort

Each decision is assessed based on risk, feasibility, and cost. We compute a risk-adjusted expected cost:

$$\text{Expected Cost} = C_{\text{direct}} + C_{\text{indirect}} + P_{\text{failure}} · C_{\text{failure}}$$

This allows the system to recommend actions that balance safety, mission success, and operational constraints.

How We Built It

ARTEMIS-Q is a modular full-stack system:

Frontend: React and TypeScript (Vite) for interactive mission visualization Core Logic: TypeScript modules for optimization, simulation, and decision-making

Key components include:

optimizer.ts for QUBO-based trajectory planning monteCarlo.ts for uncertainty simulation crewRisk.ts for radiation and health modeling missionDecision.ts for decision logic replanCost.ts for economic analysis

The architecture is designed to be extensible, allowing additional constraints and models to be integrated.

Challenges We Ran Into

Balancing realism with feasibility was a major challenge. We needed models that were meaningful but computationally tractable within the scope of a hackathon.

Integrating multiple domains, including optimization, human health, and economic modeling, required careful abstraction to keep the system coherent.

Modeling uncertainty in a way that meaningfully influenced decisions, rather than simply adding noise, was also non-trivial.

Finally, making decisions interpretable was critical. The system needed to explain why a mission should proceed, not just output a result.

What We Learned

We learned that mission planning is fundamentally about tradeoffs under uncertainty rather than finding a single optimal solution.

Incorporating human factors such as radiation exposure significantly changes how optimization problems must be structured.

We also saw the importance of combining optimization, simulation, and decision-making into a unified system, rather than treating them as separate steps.

What’s Next for ARTEMIS-Q

Future work includes:

Incorporating more realistic orbital mechanics models
Improving biomedical modeling for crew health
Integrating real-time telemetry inputs
Exploring reinforcement learning for adaptive mission decisions
Final Thought

ARTEMIS-Q is not just about optimizing where a spacecraft should go. It is about determining whether a mission should proceed at all under uncertainty, risk, and human constraints.