Nonlinear Peer-Aware Cost Estimation

Research project on nonlinear cost-driven robot navigation under real-time sensing constraints.

$cat case-study.md

Summary

Aug 2025 - Ongoing | GHOST Lab, ASU

Built a control stack where a mobile robot tracks behavior generated by a nonlinear peer-aware objective.

What I built

Motion-capture calibration for real-time rigid body localization.
ROS nodes for control and feedback logging.
EKF-based filtering for noisy mocap measurements.
MPC loop for nonlinear trajectory behavior.
Bayesian intent inference integration for belief-aware decisions.

Results

Stable closed-loop control with filtered state estimates.
Improved tracking reliability relative to raw sensor-only operation.
An IJRR manuscript submission is currently in progress.
Additional quantitative metrics are being finalized.

github back-to-projects demo (todo)