MPC Controller Skill
Overview
Expert skill for designing, implementing, and tuning Model Predictive Controllers for robotic systems, including both linear and nonlinear MPC.
Capabilities
- Derive kinematic and dynamic robot models
- Formulate MPC optimization problems (QP, NLP)
- Configure CasADi for symbolic differentiation
- Set up ACADO code generation for real-time MPC
- Implement constraint handling (velocity, acceleration, collision)
- Configure cost function weights (tracking, control effort)
- Implement warm starting for fast convergence
- Set up NMPC for nonlinear systems
- Configure terminal constraints and costs
- Optimize solver parameters for real-time execution
Target Processes
- mpc-controller-design.js
- trajectory-optimization.js
- dynamic-obstacle-avoidance.js
- path-planning-algorithm.js
Dependencies
- CasADi
- ACADO Toolkit
- OSQP
- qpOASES
- Ipopt
Usage Context
This skill is invoked when processes require advanced model-based control, trajectory tracking with constraints, or real-time optimization-based control strategies.
Output Artifacts
- MPC formulation code
- CasADi symbolic models
- ACADO generated code
- QP/NLP solver configurations
- Cost function tuning parameters
- Constraint specifications