Initial Testing

As mentioned in the previous update, the manhattan3500 data-set is being utilized to test the BMM DP– this pose-graph is depicted in Figure 1. Initially, we process the fault-free pose-graph and the pose-graph that has 100 false constraints added with traditional $L_2$ optimization.

Fig 1 :: Manhattan 3500 pose-graph

The residual scatter plots for the fault-free graph and the graph with 100 false constraints using $L_2$ optimization are shown in Figures 2 and 3, respectively. This plots depict what would be expected; both distributions have approximately zero mean with the faulty graph residuals have a larger variance.

Fig 2 :: Fault free residual scatter

Fig 3 :: Pose graph residual scatter when 100 faulty constraints are added

Using the faulty residuals, the BMM DP was tested. The BMM DP provided an inlier/outlier distribution mixture as shown below.

Now, we can test the estimated mixture by feeding it into the Max-Mixture model to see how well it optimizes the pose-graph in comparison to $L_2$.

From this table, we can see that the BMM-BP estimate mixture preforms very well in comparison to $L_2$.

Next Steps

Testing, testing, and more testing .. .