Abdullah Al Redwan Newaz

BOWConnect: Parallel Bayesian Optimization over Windows with Learned Local Cost Maps for Sample-Efficient Kinodynamic Motion Planning

Sourav Raxit, Abdullah Al Redwan Newaz, Jose Fuentes, Leonardo Bobadilla

This paper presents BOWConnect, a bidirectional parallel kinodynamic motion planner that addresses three fundamental limitations of existing sampling-based methods: sample inefficiency in high-dimensional state spaces, unreliable cost heuristics under dynamic constraints, and poo…Read more