#include <recognition_probability_computer.h>
Public Member Functions | |
| virtual void | computeRepresentationProbabilities (std::vector< ObjectRepresentation > &objects) |
| LearnedProbabilityComputer () | |
Private Types | |
| enum | Cases { CLUSTER_ONLY, CLUSTER_DB, DB_ONLY, DB_MULTIPLE, CLUSTER_DB_MULTIPLE } |
Detailed Description
Computer which assigns static weights to:
- The first detected DB model
- The cluster planner
- All of the rest of the DB models
CASES: -If there is only the cluster, the cluster gets 1.0 -If there is the cluster and one DB model, the DB model gets 0.667 and the cluster gets 0.333 -If there is only a single DB model, the DB model gets 100% -If there are only DB models, the first gets 0.667 and the rest get 0.333 divided evenly -If there are multiple DB models and the cluster, the first DB model gets 0.5, the cluster gets 0.25, and the rest get 0.25 divided evenly
Definition at line 131 of file recognition_probability_computer.h.
Member Enumeration Documentation
Definition at line 134 of file recognition_probability_computer.h.
Constructor & Destructor Documentation
Definition at line 136 of file recognition_probability_computer.h.
Member Function Documentation
| void probabilistic_grasp_planner::LearnedProbabilityComputer::computeRepresentationProbabilities | ( | std::vector< ObjectRepresentation > & | objects | ) | [virtual] |
Given a list of GraspableObjects (database models, point clouds, etc.) which describe the real object, computes the probability for each object that the hypothesis "this 'object' is the best description of the real object" is true.
Implements probabilistic_grasp_planner::RecognitionProbabilityComputer.
Definition at line 162 of file recognition_probability_computer.cpp.
The documentation for this class was generated from the following files: