PoLAR: Factorizing Extent and Mode in Latent Actions for Robot Policy Learning
PoLAR introduces Polar Latent Actions with Radial structure to better represent robot actions by separating transition extent and mode. It uses temporal offsets between observations as a proxy for extent, mapping larger gaps to larger radii in hyperbolic space. This structured latent action pretraining significantly improves downstream robot policy performance in both simulation and real-world tasks.