Akule
arXiv 2026 · Under review at AAMAS 2027 Fast and scalable multi-robot motion planning through sparse interaction diffusion, learned pairwise selection, and context-aware signed composition, with planning for up to 256 robots.
Selected research systems
Research systems & artifacts.
A focused selection of research projects, paper implementations, and technical systems.
arXiv 2026 · Under review at AAMAS 2027
SCoUT
RLC 2026 · OralScalable communication for multi-agent reinforcement learning through utility-guided temporal grouping and counterfactual credit assignment.
Massive budgeted POMDPs
TMLR 2026Oracle-guided meta-reinforcement learning for massive multi-component maintenance and inspection problems.
Risk-aware planning
Research projectReachability-based trajectory design that balances path length and risk around obstacles.
Deep RL for Intelligent Traffic Light Control
Technical projectComparison of DQN and PPO policies for intelligent signal control in a SUMO-simulated traffic intersection.
Vision-based Spacecraft Trajectory Estimation Using Structure from Motion
Technical project3D reconstruction of asteroid Bennu’s surface and spacecraft trajectory estimation using structure-from-motion techniques.