The Flow has not summarised this story yet — read it at arXiv Machine Learning.
The paper introduces Reachability Analysis-Informed Reinforcement Learning (RARL) for designing deterministic multi‑impulse interplanetary transfers. RARL uses local first‑order reachability maps to bound velocity perturbations and selects intermediate waypoints, which are then translated into maneuvers via Lambert reconstruction and a terminal two‑impulse solution. Numerical experiments on an Earth‑Mars benchmark show that RARL achieves a mean maneuver cost only 1.72% above a validated convex programming reference and can be trained once to handle a wide range of departure states, achieving 100% feasibility on 10,000 held‑out Monte Carlo departures.
This paper presents a distribution-agnostic robust trajectory-optimization framework based on chance-constrained reinforcement learning. The uncertainty is represented here through initial conditions and process noise, with the only requirement being that it can be sampled.
The paper introduces Action Diffusion, a conditional diffusion model that generates action sequences for a point-mass system affected by dry friction and stiction. Using a compact 1D U‑Net, the model produces bounded control sequences conditioned on initial and target states, outperforming uniform random shooting, structured random shooting, and the Cross‑Entropy Method in reducing terminal error and stuck steps, especially with few samples. The results demonstrate that conditional diffusion can produce temporally coherent controls that effectively overcome stiction by recombining structured primitives from the training prior.
arXiv:2606. 15359v1 Announce Type: new Abstract: Diffusion models have emerged as powerful tools for planning and control by learning multimodal distributions over actions and trajectories.
arXiv:2607. 24051v1 Announce Type: cross Abstract: Low-thrust trajectory optimization is a core technology in deep-space mission design.
arXiv:2606. 11019v1 Announce Type: cross Abstract: Learning-based motion planners, despite recent progress, often suffer from temporal inconsistency.