Track manipulation, locomotion and embodied systems.
Search papers, labs, and topics across Lattice.
Informed Belief Localization Trees* (Informed BLT*) is a novel sampling-based belief space planning algorithm designed for large-scale digital twins using point-cloud observations. By adapting RRT* and Informed RRT* with the $2$-Wasserstein metric and assuming isotropic Gaussian beliefs, it efficiently connects belief states while incorporating information and probabilistic collision constraints. This approach accelerates initial solution discovery and achieves competitive cost convergence in complex simulated and digital twin environments.
Planning in massive digital twins with point-cloud data becomes significantly faster by connecting belief states using a $W_2$-metric adaptation of RRT*, reusing measurement information to avoid repeated propagation.
We present Informed Belief Localization Trees* (Informed BLT*), a sampling-based belief space planning (BSP) algorithm that scales to large outdoor digital twins with point-cloud observations. We adapt RRT* and Informed RRT* to belief space using the $2$-Wasserstein ($W_2$) metric. Assuming isotropic Gaussian beliefs, sampled belief states can be connected efficiently while accounting for available information and probabilistic collision constraints. This enables steering and rewiring without repeatedly propagating observations, and allows previously computed measurement information to be reused. We present a framework to generate semantically labelled digital twins for planning in real-world environments with point-cloud-based localization. Experiments in simulated environments and digital twins show faster initial solution discovery in most maps with competitive cost convergence.