control Directory Reference

Directory dependency graph for control:

Detailed Description

Robust DDP Controller Documentation

This directory contains documentation for the robust Differential Dynamic Programming (DDP) controller for engine thrust regulation.

Overview

The robust DDP controller is a complete closed-loop control system that:

• Regulates engine thrust via pressure regulation
• Maintains mixture ratio constraints
• Handles model uncertainty and disturbances
• Provides safety filtering and robustness guarantees

Documentation Structure

• README.md (this file) - Overview and navigation
• DYNAMICS.md - Discrete-time dynamics model
• ENGINE_WRAPPER.md - Engine physics wrapper and caching
• CONSTRAINTS.md - Constraint checking and margins
• ROBUSTNESS.md - Residual bounds and tube propagation
• DDP_SOLVER.md - DDP optimization algorithm
• REFERENCE.md - Reference trajectory generation
• ACTUATION.md - Actuation command generation
• SAFETY_FILTER.md - Safety filtering and reachable tubes

Quick Start

See the main controller class: engine.control.robust_ddp.controller.RobustDDPController

from engine.control.robust_ddp import RobustDDPController, ControllerConfig
from engine.pipeline.config_schemas import PintleEngineConfig
 
# Initialize
cfg = ControllerConfig(...)
engine_config = PintleEngineConfig(...)
controller = RobustDDPController(cfg, engine_config)
 
# Control loop
for step in range(num_steps):
    actuation_cmd, diagnostics = controller.step(meas, nav, cmd)
    # Apply actuation_cmd to hardware

Integration

The controller integrates with:

• Engine Physics: Uses PintleEngineRunner via EngineWrapper
• Dynamics Model: Discrete-time state-space model
• Safety Systems: Constraint checking and safety filtering
• Logging: Structured logging for analysis

Frontend Integration

The controller is integrated into the web UI with flight-simulation coupling:

• Backend (backend/routers/control.py): POST /api/control/init, /step, /simulate, /reset, and GET /api/control/status.
• Frontend (frontend/src/components/ControllerMode.tsx): command-mode selection (target thrust or altitude goal), initial-condition entry, and real-time visualization of thrust tracking, pressures, actuation duty cycles, mixture ratio, value function, control effort, ullage volumes, and mass flow rates.

Each controller step calls engine_wrapper.estimate_from_pressures(...) → PintleEngineRunner.evaluate(...) for full engine physics, propagates the dynamics state, and (in /simulate) uses the resulting thrust to update altitude/velocity for closed-loop flight coupling.

See Also

• Tests: tests/control/robust_ddp/
• Code: engine/control/robust_ddp/
• Analysis Tools: tools/analyze_controller_run.py