1: \begin{abstract}

2: \ac{GP} formulation of continuous-time trajectory offers a fast solution to the motion planning problem via probabilistic inference on factor graph. 

3: However, often the solution converges to in-feasible local minima and the planned trajectory is not collision-free. 

4: We propose a message passing algorithm that is more sensitive to obstacles with fast convergence time.

5: We leverage the utility of min-sum message passing algorithm that performs local computations at each node to solve the inference problem on factor graph.

6: We first introduce the notion of compound factor node to transform the factor graph to a linearly structured graph.

7: We next develop an algorithm denoted as \ac{MS2MP} that combines numerical optimization with message passing to find collision-free trajectories.

8: \ac{MS2MP} performs numerical optimization to solve non-linear least square minimization problem at each compound factor node and then exploits the linear structure of factor graph  to compute the \ac{MAP} estimation of complete graph by passing messages among graph nodes.

9: The decentralized optimization approach of each compound node increases sensitivity towards avoiding obstacles for harder planning problems. 

10: We evaluate our algorithm by performing extensive experiments for exemplary motion planning tasks for a robot manipulator.

11: Our evaluation reveals that \ac{MS2MP} improves existing work in convergence time and success rate.

12: \end{abstract}

13: