Learning Adaptive Solvers for Distributed Factor Graph Optimization on Matrix Lie Groups
This work addresses the need for adaptive distributed solvers in large-scale robotic perception, reducing reliance on hand tuning for optimization over matrix Lie groups.
DeepCORD introduces a learning-augmented framework for distributed factor graph optimization on matrix Lie groups, achieving lower objective values than existing distributed baselines on most benchmarks for SE(3) pose graph optimization and SL(4) projective submap alignment.
Modern robotic perception increasingly involves large-scale geometric optimization problems distributed across multiple robots or sessions. However, existing distributed solvers often depend on brittle hand tuning and primarily target rigid body pose graphs. To address this, we present DeepCORD, a learning-augmented framework for distributed factor graph optimization on general matrix Lie groups. By unfolding a parallel and accelerated Riemannian optimizer into differentiable iterations, DeepCORD learns a self-supervised feedback policy that dynamically adapts solver parameters according to the optimization phase and communication status. The resulting method enables adaptive distributed optimization over matrix Lie groups under both synchronous and asynchronous communication regimes. Extensive experiments on real-world $\mathrm{SE}$(3) pose graph optimization and $\mathrm{SL}$(4) projective submap alignment show that our method achieves lower objective values than existing distributed baselines on most benchmarks across realistic operating scenarios.