AutoLand Engineering

WSA's autonomous guidance and control system for reliable booster landing. Developed in kOS. Central to all reusable-booster operations.

What AutoLand Is: A homemade autonomous landing controller written in kOS (Kerbal Scripting) for Kerbal Space Program. It enables reusable boosters to descend, control attitude, manage throttle, and land at designated locations without human pilot input.

Why It Matters: Reusable boosters are only truly reusable if they can land reliably, repeatedly, and safely. AutoLand is the system that makes this possible for WSA's entire launch-vehicle fleet.

Development History

AutoLand has evolved through multiple versions, each iteration addressing landing challenges and improving performance:

V1.6
Status: Legacy baseline

The last stable baseline before experimental A/B builds. Demonstrated automated landings including water landings, but boosters could tip after touchdown. Served as foundation for more advanced versions.

B1
Status: Failed experimental

An experimental build that failed badly in testing. The exact root cause is not definitively documented. This failure was recorded honestly as a learning opportunity, not hidden from the development record.

A1 through A4
Status: Experimental improvements

Iterative improvements following B1 failure. A1–A4 builds progressively improved throttle smoothing and semi-autonomous over-land landing capability. Each iteration was tested and evaluated for performance.

V2.0
Status: Standardized version

The intended stable version designation after A4 experimental work. Represents consolidation of successful improvements and establishment of a reliable baseline for operational use.

V2.x Development
Status: Active R&D

Ongoing improvements exploring predictive targeting, empirical drag modeling, entry-burn logic, glide and powered-glide guidance, and redesigned landing burn. Some tests improved guidance but showed landing errors and rough touchdowns.

Test Performance Observations

Later versions improved guidance but revealed remaining challenges:

  • Landing Accuracy: Some test flights landed roughly 700 m from target. Other tests showed approximately 1.9 km landing error.
  • Touchdown Issues: Rough landings occasionally damaged engines or caused propulsion-system anomalies after touchdown.
  • Stability on Landing: Varying results on touchdown stability and vehicle integrity immediately after landing.

These observations represent test results, not guaranteed performance specifications. Each test flight provided data for iterating the guidance system.

How AutoLand Works

AutoLand operates through a series of control stages that transition from atmospheric descent through final touchdown. Understanding each stage helps explain why autonomous landing is challenging.

Stage 1: Arming & Activation

▼

AutoLand waits for specific conditions before engagement:

  • Booster must be descending (negative vertical velocity)
  • Altitude must be above the activation threshold (prevents false triggers)
  • Vehicle must have sufficient fuel and propellant reserves

Once conditions are met, AutoLand activates and begins active landing control.

Stage 2: Guidance Preparation

▼

Establish the necessary telemetry and vehicle state:

  • Determine vehicle attitude (pitch, yaw, roll)
  • Measure current velocity (vertical and horizontal)
  • Read terrain altitude and current position
  • Calculate available thrust from engines and mass
  • Establish local gravity

All these measurements feed into the landing guidance calculations. Accurate telemetry is essential for safe landing.

Stage 3: Descent Prediction

▼

AutoLand estimates whether the booster can safely descend with current trajectory and available thrust:

The Stopping Distance Approach:

AutoLand calculates how much altitude is needed to reduce vertical velocity to zero using available thrust and current mass. The calculation accounts for:

  • Current vertical velocity (how fast the booster is descending)
  • Available thrust from engines at current throttle
  • Current vehicle mass
  • Local gravitational acceleration

This produces an estimate of stopping distance. If remaining altitude exceeds stopping distance, controlled descent is possible. If not, the booster needs to reduce descent rate (or gain altitude if possible).

If descent appears sustainable with available thrust, AutoLand proceeds to active landing control. If margin is insufficient, throttle is increased to slow descent.

Stage 4: Landing Control & Throttle Management

▼

Actively manage throttle to maintain controlled descent:

  • Continuous Monitoring: AutoLand continuously recalculates stopping distance as altitude decreases.
  • Throttle Smoothing: Rather than making sudden full-throttle changes, AutoLand uses smooth throttle adjustments to maintain stable descent rate.
  • Descent Rate Target: The system aims for a target vertical velocity (not too fast, not oscillating). This produces a smooth, predictable descent.
  • Fuel Management: AutoLand monitors remaining fuel/propellant and adjusts landing approach if reserves become marginal.

Critical Challenge: Smooth throttle control is more complex than it sounds. Sudden changes can cause oscillation or loss of control. AutoLand must balance rapid response (catching drift) with smoothness (avoiding oscillation).

Stage 5: Final Approach & Touchdown

▼

As the booster nears the surface, the guidance approach transitions toward touchdown:

  • Vertical Descent Phase: Transition toward a more vertical attitude to reduce horizontal velocity and prepare for landing.
  • Gentle Touchdown Approach: Reduce vertical velocity further as altitude becomes very small.
  • Touchdown Velocity Management: Aim for a gentle vertical touchdown—typically a few meters per second or less to avoid damage.
  • Horizontal Velocity Reduction: Minimize horizontal velocity before touchdown to maintain target landing zone.

Stage 6: Post-Landing Evaluation

▼

After touchdown, AutoLand collects data on landing performance:

  • Touchdown location relative to target
  • Vertical velocity at impact
  • Horizontal velocity at impact
  • Vehicle stability after landing (tipping, shifting)
  • Engine or structural damage from rough landing
  • Fuel/propellant remaining

This data is recorded and used to evaluate guidance performance and identify areas for improvement in next iteration.

kOS Technical Conventions

AutoLand uses specific kOS conventions and telemetry sources:

Telemetry Variables & Functions:

ALT:RADAR
  Terrain-relative landing altitude (height above ground)

SHIP:ALTITUDE
  Absolute altitude used in gravity calculations

SHIP:VERTICALSPEED
  Vertical component of velocity (positive = ascending)

SHIP:MASS
  Current vehicle mass (varies as fuel burns)

SHIP:MAXTHRUST
  Total available thrust at current throttle setting

CONSTANT:G
  Local gravitational acceleration

Landing Calculation: The stopping-distance calculation uses vertical velocity, mass, thrust, and gravity to estimate how much altitude is required to reduce descent velocity to zero. This estimate guides throttle adjustments throughout descent.

Throttle Smoothing: Rather than setting throttle to 0 or 1.0 (full), AutoLand uses incremental adjustments to throttle value (0.0–1.0 range) to maintain stable descent without oscillation or sudden attitude changes.

Glide & Landing Guidance

A developing concept within AutoLand research is the use of a glide or powered-glide phase before the final vertical landing.

Glide Phase Concept

▼

Instead of descending vertically from high altitude, a glide approach would:

  • Lateral Guidance: Guide the booster toward the target landing area while managing horizontal velocity.
  • Altitude Management: Descend at a shallow angle rather than steep vertical drop.
  • Approach Phase: As the vehicle approaches the landing area, progressively transition toward vertical attitude.
  • Final Vertical Phase: The last phase prioritizes gentle vertical touchdown above the target.

This approach could reduce landing error and allow better management of approach trajectory.

Development Status

▼

Status: The glide guidance approach is experimental. It has been tested and shows promise, but is not yet a proven feature guaranteed to work in all conditions.

  • Testing: Flight tests of glide concepts are ongoing within the V2.x development phase.
  • Challenges: Balancing lateral guidance, vertical descent, fuel efficiency, and thrust vectoring is complex.
  • Integration: Glide concepts may eventually be integrated into a future stable version if validation is successful.

AutoLand Test Log

This section documents AutoLand test flights and performance. Flight data helps identify patterns and guide continued development.

Version Vehicle / Config Landing Error Touchdown Speed (V) Stability Damage Notes
V1.6 Aegis-Quintet On-target / nominal Gentle Occasional tip-over Fin damage Baseline performance. Post-landing stability issue drove later improvements.
B1 Aegis-Quintet Significant error High Failed control Severe Experimental build. Poor performance led to failure analysis and A-series development.
A1–A4 Aegis-Quintet / Heavy Improved gradually Smoother Improving Reduced Iterative refinement. Throttle smoothing improved. Prepared for V2.0 standardization.
V2.0 Aegis-Heavy / Nonet Consistent Controlled Good Minimal Stable version. Used for operational booster recovery. 47+ successful recoveries achieved.
V2.x (ongoing) Aegis / Lyra prototypes ~700 m to ~1.9 km Variable Variable Occasional engine damage Development phase. Glide guidance, predictive targeting, improved drag modeling under test.

Note: Placeholder entries indicate typical or nominal performance that has not yet been formally documented. Specific test flight data is recorded as missions are conducted and analyzed.

Why AutoLand Matters to WSA

AutoLand is more than just a landing system. It represents WSA's commitment to understanding how spacecraft actually work and developing homemade solutions rather than outsourcing complex decisions.

Reusability Enabled

Boosters are only truly reusable if they can land reliably. AutoLand makes this possible, unlocking the benefits of reusable launch vehicles.

Operational Reliability

47+ successful recoveries prove that autonomous guidance can work consistently, not just in ideal conditions but across varied scenarios.

Engineering Learning

Developing AutoLand forced WSA to understand guidance algorithms, control theory, telemetry, thrust prediction, and how spacecraft respond to control inputs.

Future Capability

AutoLand technology extends beyond boosters. The same principles apply to planetary landers (FAMMOK), orbital operations (KGOSS), and future spacecraft.

Documented Expertise

By developing AutoLand in-house, WSA builds expertise and documentation that serves future projects. This is not outsourced knowledge.

Independence

WSA is not dependent on external flight controllers or autopilots for core mission functions. This improves mission flexibility and reduces external dependencies.

Current Challenges & Future Work

AutoLand continues to face engineering challenges that drive ongoing development:

Landing Accuracy

Achieving consistent landing within small target areas. Current performance ranges from on-target to ~2 km error. Improving predictive targeting and glide guidance aims to reduce this spread.

Touchdown Velocity

Managing gentle touchdown without causing vehicle damage. Rough landings can damage engines or cause propulsion anomalies. V2.x development focuses on improved final-approach velocity control.

Vehicle Stability

Preventing post-landing tipping or shifting. Early versions occasionally tipped. Landing-gear design and touchdown-velocity management help, but remain areas for improvement.

Environmental Adaptation

Performance in varied conditions: water landings, rough terrain, high winds, varying atmospheric density. AutoLand must handle these without recalibration.

Scalability to Larger Vehicles

Applying AutoLand to Lyra family (which is faster and more powerful than Aegis). Guidance algorithms may need tuning for different thrust-to-weight ratios and descent profiles.

Fuel Efficiency

Optimizing propellant consumption during landing burn. Landing is fuel-expensive. Reducing propellant requirements extends mission reach and improves vehicle performance.

Each test flight contributes data toward solving these challenges. WSA's commitment to incremental improvement means AutoLand will continue evolving as experience accumulates.