trajectory optimization on Lie group
YeeKal
โข
โข
"#"
2023-RSS-Convex Geometric Motion Planning on Lie Groups via Moment Relaxation
Background
- local search methods: fast, but depends on good initialization
- sampling-based methods: only probabilistic complete
- global optimization:
- (discrete)Mixed-integer programming
- (continuous)Polynomial programming
Global optimization does not scale well: - robot models - problem structure
This methods: - geometric formulation for robot dynamics - exploit the problem sparsity - leverage the Lasserre's hierarchy of moment relaxation
for kinodynamic motion planning
Method
- Polynomial optimization problem(POP)
- Polynomial objective and constraints
book: An introduction to polynomial and semi-algebraic optimization