In recent years, the exploration of Mars has advanced significantly with the successful flight of Ingenuity, demonstrating the feasibility of aerial vehicles on the Red Planet. However, future missions demand more capable drones, such as hexacopters, which offer greater payload capacity, redundancy, and operational flexibility. A key challenge in designing these Mars quadrotor drones lies in the harsh Martian atmospheric environment, characterized by extremely low density and high variability. This variability introduces uncertainties that can severely impact rotor performance, leading to increased power consumption and reduced mission endurance. To address this, we focus on robust aerodynamic design and optimization of rotor shapes for Mars quadrotor drones, ensuring reliable operation across diverse atmospheric conditions. In this article, we explore the Martian atmosphere’s uncertainties, develop high-fidelity aerodynamic models using advanced computational methods, and implement robust optimization techniques to derive rotor designs that balance performance and stability. By leveraging probabilistic models and multi-objective optimization, we aim to enhance the capabilities of Mars quadrotor drones for extended exploration missions.
The Martian atmosphere is notoriously thin, with an average density of approximately 0.017 kg/m³, which is about 1% of Earth’s atmospheric density. This low density, combined with a low heat capacity, results in significant temperature fluctuations that directly affect atmospheric pressure and density. Over a Martian year, these variations can exceed 30%, posing a major challenge for rotorcraft design. For instance, during Martian summer, densities can drop as low as 0.013 kg/m³, forcing rotors to operate at higher speeds to maintain lift, thereby increasing power demands and risking system failures. To quantify this uncertainty, we analyze data from the Mars Climate Database (MCD) and surface measurements, establishing a probabilistic model for atmospheric density. This model captures the range and likelihood of density values, which is crucial for robust design. The probability distribution of Martian atmospheric density throughout the year is summarized in Table 1.
| Density (kg/m³) | Probability (%) |
|---|---|
| 0.013 | 16.67 |
| 0.014 | 10.42 |
| 0.015 | 14.58 |
| 0.016 | 39.58 |
| 0.017 | 10.42 |
| 0.018 | 8.33 |
This variability underscores the need for robust design approaches that account for multiple operating points rather than a single design condition. For Mars quadrotor drones, this means optimizing rotor shapes to minimize power consumption across the entire density spectrum while reducing sensitivity to changes. The impact of density on rotor performance can be expressed through power equations, where power P is a function of density ρ and design parameters. For example, the hover power for a rotor can be approximated as:
$$ P_{\text{hover}} \propto \frac{T^{3/2}}{\sqrt{\rho A}} $$
where T is thrust, and A is rotor disk area. As density decreases, power increases substantially, highlighting the importance of robust optimization.
To accurately model rotor aerodynamics in the Martian environment, we employ the Viscous Vortex Particle Method (VVPM), which is well-suited for low-Reynolds-number and high-Mach-number flows typical of Mars. This method avoids numerical dissipation and mesh generation issues associated with traditional CFD, making it efficient for optimization loops. The VVPM is based on the Helmholtz decomposition of velocity fields and the Lagrangian description of the incompressible Navier-Stokes equations. The velocity field u(r,t) is given by:
$$ \mathbf{u}(\mathbf{r}, t) = \nabla \phi(\mathbf{r}, t) + \nabla \times \boldsymbol{\psi}(\mathbf{r}, t) $$
where ϕ is the scalar potential and ψ is the vector potential. The vorticity dynamics are described by:
$$ \frac{D \boldsymbol{\omega}}{D t} = \boldsymbol{\omega} \cdot \nabla \mathbf{u} + \nu \nabla^2 \boldsymbol{\omega} $$
with ν being the kinematic viscosity. Discretizing the vorticity field into particles allows efficient computation of wake interactions. For rotor modeling, we couple VVPM with a second-order lifting-line theory, incorporating airfoil data via C81 tables to account for viscous and compressibility effects. The aerodynamic forces on a blade element are computed as:
$$ dL = \frac{1}{2} \rho W^2 c C_l dr_b $$
$$ dD = \frac{1}{2} \rho W^2 c C_d dr_b $$
where W is the relative velocity, c is chord length, and C_l and C_d are lift and drag coefficients from airfoil data. Thrust and power are integrated along the blade span and over all blades. This model has been validated against experimental data in low-pressure chambers, confirming its accuracy for Mars conditions. The computational workflow involves iterative solving for induced velocities and vortex shedding, enabling detailed performance analysis for various rotor shapes.
The design optimization of rotor shapes for Mars quadrotor drones involves defining design variables, constraints, and objectives that incorporate atmospheric uncertainties. We focus on optimizing chord length and twist distributions along the blade span, represented using B-spline curves. This allows smooth and local modifications. For a B-spline curve of degree k with control points A_i, the curve B(u) is:
$$ B(u) = \sum_{i=0}^{n} A_i N_{i,k}(u) $$
where N_{i,k}(u) are basis functions. For chord distribution c(r_b) and twist distribution θ(r_b), we use 4th-degree and 3rd-degree B-splines, respectively, with control points as design variables. The design vector is:
$$ \mathbf{D} = [h_0, h_1, \dots, h_4, t_0, t_1, \dots, t_3] $$
where h_i and t_i are parameters for chord and twist control points. Constraints include meeting thrust requirements for hover and forward flight across all density scenarios. For a Mars quadrotor drone with mass m and gravity g_Mars = 3.71 m/s², the hover thrust per rotor is:
$$ T_v = \frac{mg_{\text{Mars}}}{6} $$
In forward flight at 30 m/s, the thrust must balance drag, which varies with density. Additionally, we impose a maximum thrust constraint for maneuverability: T_{\text{max,hover}} ≥ 1.3 mg_{\text{Mars}}/6. These constraints ensure the rotor can operate under all expected conditions.
The optimization objectives are defined to minimize both the mean power consumption over the Martian year and the peak power required at the lowest density. This dual approach enhances robustness. The mean power μ is the expected value of power over the density distribution:
$$ \mu = E[\bar{P}] = \sum_{i=1}^{N_\rho} \bar{P}(\mathbf{D}, \rho_i) p(\rho_i) $$
where ρ_i are density values from Table 1, p(ρ_i) their probabilities, and \bar{P} is the mission-averaged power per rotor. For a typical mission profile involving hover and cruise, \bar{P} is computed as:
$$ \bar{P} = 0.84 P_{\text{hover}} + 0.16 P_{\text{cruise}} $$
The peak power P_{\text{max,hover}} is the maximum hover power at the lowest density (0.013 kg/m³) when producing maximum thrust. Minimizing both μ and P_{\text{max,hover}} reduces overall energy use and sensitivity to density variations. Thus, the multi-objective optimization problem is:
$$ \min_{\mathbf{D}} \{ \mu, P_{\text{max,hover}} \} $$
subject to the constraints above. We solve this using the Non-dominated Sorting Genetic Algorithm II (NSGA-II), a popular evolutionary algorithm for multi-objective optimization. NSGA-II employs fast non-dominated sorting, crowding distance computation, and elitism to efficiently explore the design space and converge to Pareto-optimal solutions. The optimization process involves nested loops: an inner loop for aerodynamic performance calculation via Newton iteration, a middle loop for objective evaluation across densities, and an outer loop for NSGA-II-based search. This framework allows us to generate a set of optimal rotor designs that trade off between mean and peak power.
Results from the optimization yield a Pareto front of solutions, each representing a different rotor shape. We analyze three representative designs: one minimizing peak power, one minimizing mean power, and a preferred compromise solution. Their chord and twist distributions are compared in Table 2, and performance metrics are summarized in Table 3.
| Design | Chord at Root (m) | Chord at Tip (m) | Twist at Root (°) | Twist at Tip (°) | Solidity |
|---|---|---|---|---|---|
| Min Peak Power | 0.045 | 0.065 | 15.2 | 5.8 | 0.218 |
| Min Mean Power | 0.042 | 0.055 | 14.8 | 6.2 | 0.195 |
| Preferred | 0.043 | 0.060 | 15.0 | 6.0 | 0.203 |
| Design | Mean Power μ (W) | Peak Power P_max (W) | Power Sensitivity ΔP/Δρ (W per kg/m³) |
|---|---|---|---|
| Min Peak Power | 363.71 | 691.41 | 1200 |
| Min Mean Power | 347.68 | 850.78 | 1800 |
| Preferred | 351.47 | 716.93 | 1400 |
The min peak power design features higher solidity, especially at the tip, which reduces power demand at low densities but increases profile drag at high densities. The min mean power design has lower solidity, optimizing for average conditions but suffering higher peak power. The preferred design strikes a balance, with moderate solidity and improved robustness. To illustrate power variations, we compute hover and cruise power across densities using the aerodynamic model. For hover, power generally increases as density decreases, but the rate depends on design. The relationship can be expressed as:
$$ P_{\text{hover}} \approx \frac{C_T^{3/2} \rho^{1/2} \Omega^3 R^2}{\sqrt{2}} + \frac{C_P0 \rho \Omega^3 R^4}{8} $$
where C_T is thrust coefficient, C_P0 is profile power coefficient, Ω is rotational speed, and R is rotor radius. The first term represents induced power, and the second term profile power. At low densities, induced power dominates, justifying higher solidity for reduced induced losses. However, increased chord also raises profile power, leading to trade-offs. For cruise at 30 m/s, power trends are non-monotonic due to complex wake interactions and drag effects, as captured by the VVPM model. The preferred design shows relatively flat power curves, indicating low sensitivity to density changes.
Further analysis involves examining the impact of design variables on objectives. Using sensitivity derivatives, we assess how chord and twist adjustments affect mean and peak power. For instance, the derivative of mean power with respect to chord control point h_i is:
$$ \frac{\partial \mu}{\partial h_i} = \sum_{j} \frac{\partial \bar{P}(\rho_j)}{\partial h_i} p(\rho_j) $$
These derivatives guide optimization toward robust solutions. Additionally, we evaluate robustness metrics such as the coefficient of variation (CV) of power across densities:
$$ \text{CV} = \frac{\sigma_P}{\mu_P} $$
where σ_P is the standard deviation of power. Lower CV indicates better robustness. The preferred design achieves a CV of 0.15, compared to 0.22 for the min mean power design, confirming its stability. This robustness is critical for Mars quadrotor drones, as it ensures consistent performance despite atmospheric unpredictability.
In practical terms, the optimized rotor shapes enable Mars quadrotor drones to execute longer missions with reduced energy consumption. For a typical mission involving 170 seconds of hover and 33 seconds of cruise, the preferred design saves approximately 5% in total energy compared to a baseline design optimized for a single density. Over multiple flights, this translates to extended range and operational lifetime. Moreover, the lower peak power reduces demands on motors and thermal management systems, allowing for lighter and more reliable drones. These advantages are vital for future Mars exploration, where quadrotor drones may conduct aerial surveys, sample retrieval, and scouting for rovers.
To generalize our approach, we discuss applications to other multi-rotor configurations, such as quadrotor drones on Earth or other planets. The robust design methodology can be adapted by adjusting atmospheric models and constraints. For example, for Earth-based quadrotor drones operating in varying altitudes or temperatures, similar probabilistic models can be derived, and optimization can focus on minimizing power variability. The use of VVPM for aerodynamic modeling remains applicable, especially for drones with complex wake interactions. Additionally, the B-spline representation of blade shapes offers flexibility for manufacturing constraints, such as minimum thickness or material limits.
In conclusion, robust aerodynamic shape optimization is essential for Mars quadrotor drones to cope with atmospheric uncertainties. By integrating probabilistic density models, high-fidelity vortex methods, and multi-objective evolutionary algorithms, we derive rotor designs that minimize both mean and peak power consumption. The preferred design demonstrates a balanced performance, with reduced sensitivity to density changes, enhancing mission reliability and endurance. Future work may explore additional uncertainties, such as wind gusts or terrain effects, and incorporate aeroelastic considerations for lightweight blades. As Mars exploration progresses, advanced quadrotor drones will play a pivotal role, and robust design principles will ensure their success in the challenging Martian environment.
Throughout this study, we emphasize the importance of robustness in aerospace design, particularly for autonomous systems like quadrotor drones. The methods developed here can be extended to other planetary bodies or extreme environments on Earth. By continuously refining aerodynamic models and optimization techniques, we can push the boundaries of what quadrotor drones can achieve, enabling new scientific discoveries and operational capabilities. The integration of robust design into the development pipeline will be crucial for next-generation Mars quadrotor drones, fostering more ambitious exploration missions.
Finally, we note that the success of such drones relies on interdisciplinary efforts, combining aerodynamics, control systems, and mission planning. Our focus on rotor shape optimization is one piece of this puzzle, but it highlights how careful design can mitigate environmental challenges. As we look to the future, quadrotor drones on Mars will become more sophisticated, and robust aerodynamic design will remain a cornerstone of their development, ensuring they operate efficiently and reliably across the Red Planet’s diverse conditions.