Autonomous excavation adaptation to voids and heterogeneous material

Background

Rubble pile asteroids like Bennu and Ryugu have been shown to contain significant internal voids, boulders of varying size, and heterogeneous material composition. The dual bucket-wheel excavation system validated in rq-0-26 must operate autonomously for months without ground intervention. When the excavator encounters an unexpected void (sudden loss of resistance), a large boulder (sudden spike in resistance), or a material transition (e.g., from loose regolith to consolidated matrix), it must adapt in real-time to avoid damage, maintain excavation efficiency, and continue autonomous operations.

Why This Matters

Autonomous adaptation capability determines:

• Robot operational lifetime (encountering unexpected material causes damage if not managed)
• Excavation throughput consistency (stops and restarts reduce annual yield)
• Fleet reliability (each failure removes capacity from the mining operation)
• Ground intervention frequency (adaptation failures require human decision-making)
• Minimum fleet size for reliable throughput (more robust robots = fewer needed)

At 1,000+ tonnes per robot per year over a 5-year lifetime, each robot will encounter thousands of material transitions and heterogeneities. The adaptation system must handle these routinely without human input.

Key Considerations

• Void encounter could cause loss of anchoring or sudden robot movement
• Boulder encounter could stall bucket wheels, overload motors, or break teeth
• Material transitions change optimal excavation speed and depth of cut
• Sensing lag (camera, force/torque) limits reaction time at excavation speed
• Learning from one robot's experience could benefit the fleet (shared adaptation)
• Pre-mapping with ground-penetrating radar reduces but cannot eliminate surprises

Research Directions

1. Force-torque sensing requirements: Define the sensing bandwidth and resolution needed to detect material transitions within one bucket revolution, enabling real-time speed and depth adjustment.

2. Void encounter protocols: Design autonomous responses to sudden loss of cutting resistance, including anchoring system response and safe retreat procedures.

3. Boulder management strategies: Develop decision algorithms for whether to excavate around, break through, or relocate when encountering boulders larger than bucket capacity.

4. Fleet learning architecture: Design a system where excavation experience from one robot (material maps, adaptation outcomes) is shared with the fleet to improve collective performance.

5. Simulation environment development: Create a high-fidelity excavation simulator with heterogeneous asteroid models for training and validating autonomous adaptation algorithms before deployment.