BayesianVLA: Bayesian Decomposition of Vision Language Action Models via Latent Action Queries

BayesianVLA: Bayesian Decomposition of Vision Language Action Models via Latent Action Queries explores BayesianVLA enhances robot manipulation by robustly integrating language and vision through a novel Bayesian framework, overcoming generalization issues in multi-task scenarios.. Commercial viability score: 8/10 in Robotics AI.