On-Orbit Optimal Kinodynamic Planning for Low-Thrust Trajectory Maneuvers

Using RRT* with an LQR-based heuristic to efficiently design low-thrust spacecraft trajectories, addressing nonlinear dynamics and long-duration maneuvers.

Course: CMU 16-782 Planning and Decision-making for Robotics Fall 2023
Instructor: Maxim Likhachev
Authors: Juan Alvarez-Padilla, Ibrahima Sory Sow, Fausto Vega and Nayana Survana

For our project, we would like to design and generate a low-thrust trajectory for a spacecraft orbital transfer (i.e. GEO to GTO transfer maneuver)[1]. This problem presents extensive challenges as the dynamics are highly nonlinear, the spacecraft has little control authority (several orders of magnitude difference compared to traditional chemically propelled trajectory), and is typically underactuated. Furthermore, those trajectories span over weeks and months involving a significant problem size and potential numerical difficulties.

We want to leverage the cost-to-go function of linear quadratic regulation (LQR) as a cost metric to extend a particular sampling-based planner (e.g. RRT*) and compare this approach to other optimal-based sampling-based planners. In particular, we take inspiration from [2]. We plan on implementing our approach in C++ with visualization in ROS/Rviz or Meshcat.

RRT* panner with LQR cost-to-go visualization.

Check the project here.

References

[1] Tracy, Kevin, and Zac Manchester. “Low-Thrust Trajectory Optimization Using the Kustaanheimo-Stiefel Transformation”. 2021 AAS/AIAA Space Flight Mechanics Meeting.

[2] Perez, Alejandro, et al. “LQR-RRT*: Optimal sampling-based motion planning with automatically derived extension heuristics.” 2012 ICRA.